| Age | Commit message (Collapse) | Author |
|---|
|
Brings the nine commits that are actually the latest on our lineage:
isolated testnet parameters and genesis, optional 2 MB large pages for
RandomX, and five README changes. This is what "up to date" means here --
not the 1.2.2x tags, which carry higher version numbers but were cut
2026-07-30, three weeks before the line we ship, and which upstream/main
does not contain.
Three conflicts, all resolved toward upstream's structure while keeping
what this fork knows:
main.cpp -- upstream moved genesis construction into BuildGenesisBlock() so
mainnet and the new testnet can share it. Verified byte-for-byte before
taking it: MAINNET_GENESIS_TIMESTAMP, _TIME (1753315200), _NONCE (3141),
_HASH, _MERKLE, the scriptPubKey, nBits and nVersion all equal what this
tree built inline. The mainnet genesis hash is unchanged and the asserts
still guard it. Kept our warning about never editing those inputs, and kept
our note that 1753315200 is 2025-07-24, not 2026 -- upstream's comment on
that constant states the wrong year.
net.h -- DEFAULT_PORT = htons(8433) becomes MAINNET_PORT/TESTNET_PORT in
host order. Nothing is lost: net.cpp's GetDefaultPort() applies htons()
itself and is what every caller uses, so the value reaching a sockaddr is
unchanged and mainnet stays on 8433.
selftest.cpp -- the known-tests string had diverged in both directions.
Took the union: upstream's network-params alongside our payout-ledger,
rather than dropping either.
Gates: 15/15 selftests (our 14 plus upstream's network-params), consensus
constants unchanged against e4f898c except the pre-existing
MAX_BLOCK_SIZE_GEN, mainnet genesis intact, compactblock's 17 checks still
pass, money-log check ok, GNU_STACK RW.
|
|
# Conflicts:
# Makefile
|
|
std::thread throws std::system_error when pthread_create is refused
(RLIMIT_NPROC, a container's pids cgroup, memory pressure), and nothing
caught it, so the node died by std::terminate. Threads are spawned per
peer connection, so the pressure that triggers this scales with peer
count, which is attacker-influenced. Observed killing a healthy
participant worker mid-mining.
Every spawn site in the node build, and what a failure now does:
- btftunnel.cpp: both tunnel pumps are guarded. The inbound pump starts
first because it owns cleanup of the pump end and rendezvous socket;
if the second spawn is refused, shutdown() wakes the first out of its
read and it cleans up itself -- closing those descriptors from the
failure path instead would race it onto reused descriptors. Both
paths return INVALID_SOCKET, which every caller already treats as
"this connection failed".
- btfrv.cpp: the relay's accept loop drops the connection it cannot
serve and stops accepting for a moment -- longer after consecutive
failures, capped at 5s -- instead of dying at an arbitrary one. The
per-pair second pump now runs on the handling thread itself, leaving
exactly one spawn to guard and nothing running yet to race.
- main.cpp: a participant that cannot spawn its hash workers stops the
ones that did start, drops the pool connection, and returns to the
reconnect loop, the same way an ordinary disconnect does.
- randomx_pow.cpp: dataset init joins the workers it got, releases the
dataset, and stays in light mode -- the same trade the allocation
failure path above it already makes.
- rpc.cpp, net.cpp: the remaining unchecked _beginthread calls (which
report failure by value, they cannot throw) now say precisely what
did not start. The pool's accept and payout threads failing silently
left the pool running deaf -- nobody accepts, nobody gets paid, and
nothing in the log says why.
Left alone deliberately: the per-relay accept spawns (rpc.cpp) already
check, log and retry on the next poll; startup-critical spawns in
net.cpp keep failing hard; the selftest-only std::thread uses may abort
a selftest run, which is acceptable where dying over one peer is not.
Verified: full node rebuilds clean and links; pool-stratum selftest
passes; a scratch harness under RLIMIT_NPROC=1 confirms _beginthread
reports -1 and std::thread throws system_error, the premises of every
path above.
|
|
Merges upstream tag v1.2.20 (6aa3dade) into our 1.2.19+sec tree. The merge
base is bc3de30's descendant 9e39a50 -- upstream's own v1.2.19 -- so this is
an ordinary three-way merge, not a replay. Five conflict blocks in four
files, all resolved below.
No consensus change. CheckBlock, ConnectBlock, DisconnectBlock, ConnectInputs,
AcceptBlock, AcceptTransaction, GetNextWorkRequired and GetBlockValue hash
identical to v1.2.19; script.cpp and script.h are byte-identical on both
sides. main.cpp is +3 lines and main.h +2, all wallet-backend plumbing.
rpc.cpp is untouched, so the payout ledger arrives unchanged.
Makefile, VERSION. OURS, renumbered. 1.2.20+sec, for the same reason 1.2.19
carried +sec: this tree is not the upstream release of that number and the
string names every release artifact.
Makefile, AppImage recipe. OURS for the AppDir path, THEIRS for the bundled
library. Upstream's hunk reverts the icon paths to a hardcoded
/tmp/bitflash.AppDir, which is the local-user pre-placement hazard the fork
moved out of /tmp in the first place; taking their side wholesale would have
reintroduced it silently, inside a conflict about something else. The AppDir
stays $(APPDIR), and libsqlite3.so.0 joins the bundled list, which the new
backend genuinely needs.
src/Makefile, src/makefile.mingw, HEADERS. BOTH. Ours drops market.h,
theirs adds wallet_sqlite.h; the resolution does both. NODE_OBJS auto-merged
correctly in both files -- obj/wallet_sqlite.o present, obj/market.o still
gone -- and was verified rather than assumed, because that line has trapped
this fork before.
src/selftest.cpp, known-tests list. BOTH, union. Upstream adds eight
wallet-storage tests, we keep payout-ledger. The dispatch chain itself
auto-merged and still routes payout-ledger to RunPayoutLedgerSelfTest.
Hardening verified intact after merge: the TLS 1.2 floor in nostr.cpp, the
/rvrelay pin, opt-in managed Tor, strict sig3-on-onion descriptor
verification, and the marketplace removal. Upstream shipped no build logs or
__pycache__ this time; the three added files are docs/storage.md and
src/wallet_sqlite.{cpp,h}.
The SQLite backend this release introduces is NOT enabled here. An existing
wallet.dat resolves to Berkeley DB on every branch of the new selection
logic, and the conversion wizard is GUI-only and not compiled into the
headless node. The review findings that gate ever turning it on -- the
non-atomic backend marker, the connection-wide WriteHDMaster transaction,
and the encrypt-swap crash window -- are recorded on the tracker, not fixed
here.
|
|
Upstream 1.2.19 made the P2P listener wildcard-bind unconditionally because
managed Tor forwards hidden-service traffic to 127.0.0.1, which a LAN-only
bind cannot receive. That also widened every non-Tor node to listen on VPN
tunnels, docker bridges and secondary NICs. The onion endpoint is always
configured (or the managed-Tor hidden service established) before StartNode,
so the predicate is ready when the bind happens.
|
|
Merges upstream tag v1.2.19 (9e39a50) into merge/upstream-1.2.18 (991853f).
Unlike the 1.2.15->1.2.18 replay, bc3de30 is a true ancestor of our branch via
a9b4f89, so this is an ordinary three-way merge with merge-base bc3de30.
Delta: 36 files, +3265/-163, the Tor onion transport (managed Tor, signed
direct onion endpoints, onion-only dial mode) and the nHDNext restore fix.
src/main.cpp, main.h, script.cpp/h and db.cpp/h are byte-identical between the
two tags, and the poll() socket loop carries no diff hunks, so the merge is
consensus-neutral and the soak-tested socket paths are unchanged. relay/ is
untouched; the payout ledger (rpc.cpp) is untouched.
7 conflict blocks in 6 files, resolved as follows.
Makefile -- VERSION. Ours: stays 1.2.19+sec with the naming comment updated,
so artifacts from this tree cannot be mistaken for the upstream release.
src/Makefile, src/makefile.mingw -- HEADERS list. Ours (no market.h) plus
their tor.h. NODE_OBJS auto-merged correctly: no market.o, tor.o present.
src/selftest.cpp -- known-tests list. Union: payout-ledger (ours) and
managed-tor (theirs).
src/main_gui.cpp -- one block. Our /dumpaddresses kept; their
BtfStopManagedTor() on the StartNode failure path taken. The /rvrelay pin,
/noannounce and the marketplace removal auto-merged and were verified intact.
src/nostr.cpp -- two blocks in ResolveDescriptor. Signature takes both
out-params (our fExchangeOk, their rawOut). The body keeps our
drain-to-EOSE read -- returning early poisons the pooled socket for the next
caller -- and adds their rawOut capture. Our TLS 1.2 floor
(SSL_CTX_set_min_proto_version) is untouched by this merge; upstream 1.2.19
still has no floor of its own.
Upstream's release commit also added four build logs and a compiled Python
cache (build-1219.log, build-17.log, build-appimage.log, build-node.log,
scripts/__pycache__/bitflash_chain.cpython-313.pyc). Removed here; they are
build output, not source.
|
|
Merges upstream tag v1.2.18 (bc3de30) into our 24 commits. The merge base is
a259215, upstream's own commit, whose src/ tree is byte-identical to 68de575 --
our commits were replayed onto a259215 first so that git had a real base to work
from. That produced 28 conflict hunks in 9 files instead of the 21 conflict
events / 51 hunks a direct cherry-pick onto the tag would have forced, most of
which were artifacts of lifting our commits away from their own predecessors.
No consensus change. main.cpp's GetBlockValue, GetNextWorkRequired,
ConnectInputs, ConnectBlock, CheckBlock, AcceptBlock and AcceptTransaction are
untouched by this merge; script.cpp, serialize.h and randomx_pow.h are
identical on both sides. The frozen-chain oracle reproduces the pinned UTXO root
116a2b6b... byte for byte, and the merged binary loads the same 12,586-block
chain to the same tip aac622f0a3b2f8... with all blocks re-verified and none
rejected.
Which side won, per file and per hunk.
src/rpc.cpp -- 15 hunks.
1. RebuildJob transaction selection. THEIRS for the multi-pass loop and the
nFees fix, OURS for the cap. Their vfAlreadyAdded/fFoundSomething loop lands
dependent transactions that our single pass dropped deterministically, and
their nFees is accumulated across the scan instead of being redeclared inside
it -- ours passed a hardcoded 0 to GetBlockValue, so every pool block built
here discarded its fees. But their bound is MAX_SIZE/2 = 16 MB, and
MAX_BLOCK_SIZE is 1 MB, so their template can exceed what their own
CheckBlock accepts. Cap kept at MAX_BLOCK_SIZE/2, with our inner break so a
single pass cannot overshoot by more than one transaction. Our RebuildJob
sync gating and deaf-node detection are kept; upstream has none.
2. struct PendingPayout. OURS for the key, THEIRS for paidTxids. Their id is
the string "block-<height>", which is height keying -- the exact bug 44849e6
fixed after a 280-block reorg left ~14.4k BTF owed against blocks that no
longer existed. Height keying also lets the crash-recovery probe match the
previous round's transaction after a reorg and silently skip a miner, so
their id is dropped entirely and blockHash stays the key. paidTxids is taken
because it is what makes an entry survivable. Their nextAttempt and our
retryAfter are the same concept under two names; ours is kept and theirs
dropped. matureAtHeight gained an explicit = 0, which neither side had in
our style of initialiser.
3./4. Ledger writer and loader. Mechanical follow-ons of 2. The loader keeps
ours and takes their tolerance for an entry with no usable height, guarded so
it cannot fire on a well-formed entry. One redundancy was removed rather than
merged: their unguarded `if (obj.contains("blockHeight")) get<int>()` sat two
lines above our is_number-guarded version, and an unguarded get<int>() on a
malformed value throws out of the whole loop, losing every entry in the file.
Verified against the live production pending_payouts.json (2 entries, keys
amounts/blockHash/blockHeight/matureAtHeight): loads both. A file in the new
shape, with retryAfter and paidTxids, also loads.
5. Insertion point after the loader. BOTH. Our PayingRecord/gPaying in-flight
ledger and their PoolRoundProof/PoolRoundToJson are disjoint features that
landed at the same offset. Both kept, in full.
6. Writer tail. THEIRS, copied into our writer. Checking fclose() matters here
because fclose is where the buffered write actually reaches the OS: renaming
a short .tmp over a good file is precisely the truncation the atomic write
exists to prevent. The Windows remove()-before-rename() is required because
rename() there will not replace an existing file.
7. Loader bodies. BOTH -- LoadPaying and LoadPoolRounds are different functions
that happened to share a common tail.
8. Function block, 111 vs 36 lines. BOTH, with one deliberate deletion.
PayoutBlockStillOnChain, PayoutComment, RecordPoolRound, MarkPoolRoundPaid
all kept. WalletAlreadyPaid is DELETED and ReconcilePayouts rewritten,
because our reconciliation and their FindRecordedPoolPayout are two answers
to the same crash-recovery question and running both would give the same debt
two authorities. Their wallet probe wins: it matches structured metadata
rather than a comment string, so it can tell two rounds paying the same
address the same amount apart. ReconcilePayouts keeps only the job that is
still real -- dropping in-flight records the ledger already shows paid, and
arming each restored record for its one wallet probe. It no longer touches
gPendingPayouts at all; under the new model dropping ledger entries there
would reintroduce the loss this merge adopts their model to prevent. Their
gPayoutExecMutex/gPayoutThreadRunning re-declarations were dropped as
duplicates of ours, and gPayoutExecQueue was not taken: gPaying is the exec
queue in this tree, at finer granularity.
9. QueuePayouts signature. THEIRS for the round-proof arguments, OURS for the
type of blockHash. It stays a uint256 and is converted to hex only where the
round proof needs text -- a ledger key that has been through ToString() is a
key that can be compared wrong.
10. Maturity policy. OURS. matureAtHeight = blockHeight +
COINBASE_SPENDABLE_DEPTH (120), not + COINBASE_MATURITY (100). At depth 100
the block's own coinbase is still 20 blocks short of spendable, so the
payment must come from unrelated mature coin and the pool runs a permanent
float. Our own history contradicts itself here -- d6972a0 argued for
COINBASE_MATURITY citing "96% of 934 payments", 44849e6 reversed it on the
grounds that the measurement was taken while the node was alone on a private
fork paying itself. HEAD's position is kept. This is a considered
disagreement with upstream, not an oversight. The live production ledger
confirms it: both entries have matureAtHeight - blockHeight = 120.
11. Comment above FlushMaturePayouts. Rewritten to describe the merged design.
12. FlushMaturePayouts, the central difference. THEIRS for the model. The ledger
entry now survives on disk until every recipient has a txid, instead of
being erased as soon as its payments were handed off -- the old ordering
left a window where the debt was described by nothing at all, and a
container restart on 2026-07-30 silently dropped ~1,156 BTF through it. Our
gPaying is kept on top as the in-flight working set, because it carries the
per-recipient retry and deferral state a per-round nextAttempt cannot
express. It is explicitly a cache derived from the ledger, not a second
record of debt: if it were lost the next flush would rebuild it.
13. PayoutThreadFn, 79 vs 70 lines. OURS for the loop and the reorg re-check,
THEIRS for the per-recipient dedupe. The re-check on every attempt is
load-bearing and upstream has no equivalent -- it decides once, at maturity,
and a payment can sit through hours of retries during which the reorg that
kills its funding block can land. Their paidTxids check is taken, against
the ledger rather than against the batch. Our deferral policy is kept:
waiting for coin is not a failure and must not consume the retry budget.
Two fixes upstream does not have are folded in here rather than deferred to
a follow-up commit:
- FindRecordedPoolPayout runs once per record, on its first attempt, not on
every attempt. It is a linear scan of mapWallet under cs_mapWallet, per
recipient; upstream repeats it for every retry, which with 2000 rotating
payout addresses and a budget of 1440 retries is a full wallet scan per
address per minute for a day. The question it answers can only change
across a restart, and every restart re-arms the flag.
- The ledger file write no longer happens under gPayoutLedgerMutex in the
per-recipient loop. SavePendingPayoutsLocked was split into a serialise
step that needs the lock and a write step that does not, so the mutex is
held only long enough to read the structure. The orphan scan in
FlushMaturePayouts was likewise restructured so the ledger mutex is not
held across PayoutBlockStillOnChain's cs_main acquisition.
One case the merged model created and neither side handled: under
survives-until-fully-paid, our GIVING UP path dropped the in-flight record
but left the recipient owed in the ledger forever, so every later flush
would explode it again -- a forfeit that silently becomes an infinite retry.
ForfeitPayoutRecipient writes the amount off so the entry can retire.
14. QueuePayouts call site. Follow-on of 9.
15. ThreadRPCServer startup. BOTH. LoadPoolRounds added; LoadPaying and
ReconcilePayouts stay, and stay before the payout thread starts.
Crash safety, stated as the property that must hold: after kill -9 at any
point, restart pays each recipient exactly once. Before SendMoney, the
ledger has no txid and the payment is retried. Between SendMoney and the
ledger write, the wallet holds the transaction and the first-attempt probe
finds it. After the ledger write, paidTxids suppresses it. If the in-flight
file is lost entirely, the ledger rebuilds it and the probe still covers the
middle window.
src/nostr.cpp -- 3 hunks. THEIRS as the base, with two things restored from
ours. Theirs loads the Windows ROOT certificate store, which ours lacks and
genuinely needs, since OpenSSL's default verify paths are usually empty there
and our version would fail closed. But the plan's claim that theirs is a
superset of our 06bceb4 is FALSE: upstream's nostr.cpp contains no
min_proto_version, no TLS1_2, no SSL_OP_NO_*, and no set_options anywhere, so
taking it wholesale would have removed the TLS-version floor outright. Restored.
Second, their SSL_get_verify_result check sits after a successful handshake,
where with SSL_VERIFY_PEER set it is nearly dead code -- a bad certificate
aborts inside SSL_connect. The reporting belongs in the failure branch, which is
where the merge put it; their post-handshake duplicate was dropped. Our
relay-list and socket-pooling hunks auto-merged and were verified intact.
src/net.cpp -- 1 hunk. THEIRS. Their guard, hSocket == INVALID_SOCKET ||
pnode->hSocket != hSocket, subsumes ours and additionally catches descriptor
recycling, which ours did not; our version also referenced a pnode that only
existed in the select() loop. Our other db34011 hunk, nulling pnode->hSocket in
the poll-failure branch, auto-merged and was NOT re-applied -- plan section 7
Batch D says to re-apply it, and that instruction is stale. Two pieces of
residue the merge left behind were cleaned up: a comment still describing
select() semantics, and a pnode->hSocket != INVALID_SOCKET guard made redundant
by a continue two lines above it.
src/main.cpp -- 2 hunks. h1 BOTH, pure add/add adjacency between our
ParticipantHashThreadCount and their SecureClear forward declarations. h2 OURS,
and their handler discarded: 44004e3 no-ops checkorder/submitorder but keeps
market.cpp and market.h, whereas 11766be deletes the marketplace outright
including CRequestTracker, subscribe/publish and mapProducts. Ours is strictly
more.
src/main_gui.cpp -- 2 hunks. BOTH, additive on both sides. Our /payoutaddr,
/payoutfile and /hashthreads alongside their /stratumbridge and
/stratumbridgeport.
src/btfrv.cpp -- 1 hunk, not predicted by the plan. False adjacency: our
INVALID_SOCKET check and their #endif landed at the same offset. Ordering is
load-bearing -- our check goes inside the #else arm, before the #endif, or the
non-proxy build silently loses it. The BITFLASH_NODE_BUILD arm already returns
early, so both paths now have the check.
Makefile -- 2 hunks plus a hazard with no conflict marker. THEIRS for both
dependency-fetch recipes: checkout --detach FETCH_HEAD followed by test
"$(git rev-parse HEAD)" = "$(..._COMMIT)" asserts the pin actually landed, where
ours only checked it out. Everything of ours that they lack is kept: the
appimagetool 1.9.1 release pin with a sha256 check, where upstream fetches the
mutable "continuous" tag and executes it undigested, and the move of both the
tool and the AppDir out of /tmp, where any local user could pre-place an
executable and have it run as whoever runs make.
The hazard: both sides added a pin block in a different place, so git merged
both, outside any conflict region. SECP256K1_COMMIT and RANDOMX_COMMIT ended up
assigned twice, near L16 and near L44, and GNU Make takes the last := -- a naive
resolution would have silently built against our old secp256k1 pin while
appearing to adopt upstream's. The duplicate block was deleted by hand. Each of
the four variables is now assigned exactly once, and make -n resolves
SECP256K1_COMMIT to 7fecac74 with 6e2c8bc4 appearing nowhere.
Also applied here, from plan phase 4: SECP256K1_COMMIT moved to upstream's
7fecac74aed8e1fd9078380d67dd04663705c989 so signature validation matches the
binaries the rest of the network runs. secp256k1 was rebuilt from scratch
against the new pin on the build host and the checked-out dependency commit
verified. RANDOMX_COMMIT already matched at 1e9d4b2d and is unchanged.
src/Makefile, src/makefile.mingw -- 1 hunk each, both the HEADERS list. Ours
drops market.h, theirs adds proxy.h; the resolution does both. Taking theirs
wholesale would have left the Makefiles referencing a header git had already
deleted in our favour, silently, without a conflict.
Files the plan predicted would conflict and did not: net.h, main.h and db.cpp
auto-merged. Verified rather than assumed. net.h carries both our
BTF_MSG_WAIT_MAX_SECS work, including the constructor fix from 2d84982 that
starts nMsgWaitSince and nMsgWaitSize at 0, and their comment rewrite with
extern int nPeersWatched. The dead members that fix removed are gone from every
file.
Deviations from MERGE-1.2.18-PLAN.md.
- Section 6's mechanism was not used. Replayed onto the real merge base and
merged, rather than cherry-picking onto the tag. Same result, 28 hunks
instead of 51.
- Section 2 calls 09ff575 a superset of our 06bceb4. It is not; upstream has no
TLS-version floor at all. Ours restored on top of theirs.
- Section 7 Batch D says to re-apply our db34011 line into their poll-failure
branch. It auto-merged; re-applying would have duplicated it.
- Section 2 says do not take the stratum bridge. It is taken, and left off.
fStratumBridge defaults false in main.cpp, the thread is only spawned when
/stratumbridge is passed, and the listener binds 127.0.0.1 only. Excising it
would cost real divergence and future conflict for no gain while it is
disabled. Its defects -- select()/FD_SET on a node that can hold more than
1024 descriptors, and an uncapped thread per connection -- are real only when
it is enabled and are being reported upstream separately.
- Section 7's staged batches were not run as separate gated commits. The
conflict surface is interleaved within single functions, chiefly the payout
ledger, and splitting it would have meant committing states that do not
build. The merge is one commit, as briefed.
- Batch F's two fixes are folded into hunk 13 rather than deferred, because
they change the same lines the resolution rewrites.
- The plan did not anticipate the btfrv.cpp conflict, the duplicate pin block,
the market.h trap in the two Makefiles, or the give-up/infinite-retry
interaction created by adopting upstream's ledger model.
|
|
StartMinerThreads() sizes its pool with MinerThreadCount(), which is right for
solo mining -- each thread hashes a template of its own -- and wrong for
participant mode, where every thread entering BitcoinMiner() is a whole miner:
its own Stratum session, its own payout address off the rotation, its own pool
of /hashthreads workers. The two counts multiplied instead of dividing.
Measured on the pool box with /hashthreads=2: a container with no cpuset logged
"Mining with 11 thread(s) of 12 core(s)" -- 11 sessions, 22 hash threads on 12
cores, 1058% CPU, load 22.68 -- while the same binary under --cpuset-cpus 10,11
logged one session and 199%. The second miner (8 cores, /hashthreads=1) ran 7
sessions and went 74C to 89C. No code differed between those runs. Only the
cpuset did, which means the cpuset was holding the thread count down by
accident and every deployment without one was oversubscribed.
Participant mode now starts exactly one session, so /hashthreads means what it
says instead of being multiplied by the core count. The worker count moves to
ParticipantHashThreadCount(), next to MinerThreadCount() where the two are
easy to confuse, so what is printed at startup and what actually hashes cannot
drift apart -- and it IS printed now. It was only ever visible under -debug,
which is how eleven sessions ran unnoticed.
Same host, same cpuset (0-3), same arguments, measured as a delta over 20 s
more than four minutes after start:
before 399% CPU, 6-7 threads hot, 3 sessions
after 201% CPU, exactly 2 threads at 99.9%, 1 session
Solo mining is untouched: /gen without /participant still starts
MinerThreadCount() threads.
(cherry picked from commit 45bc9ed5bc132712092f3e7902fef746c6cae898)
|
|
Four inherited handlers took attacker-controlled input and wrote to disk
with nothing authorising it. None of them had a caller left anywhere in
the tree, so all four were reachable only from the outside.
checkorder called GenerateNewKey() -- AddKey(), a wallet.dat write -- for
any peer that asked, and handed back the pubkey. Its comment claims the
cache keeps one key per ip, and on a 2009 network of routable addresses
it did. Not here: inbound peers arrive through a rendezvous tunnel and
are tagged by BtfMarkerAddr() over three random bytes, so addr.ip is
fresh every connection and the cache never hits. Reconnect in a loop and
the node mints a key per connection, none of them in any backup taken
beforehand, which for a pre-BIP32 wallet is how coins go missing. The map
holding them was static and never trimmed.
submitorder AddToWallet()'d a CWalletTx from any peer that sent one. It
had to be a valid transaction; it did not have to be ours.
review checked a signature the sender made with its own key, which proves
possession and authorises nothing, then appended to reviews.dat by
reading the whole vector, pushing one entry and writing it all back --
quadratic disk on an unbounded map<string,string> off the socket -- then
walked a link graph the same peer supplied and relayed the message to
every other node.
reply dispatched into CRequestTracker. Nothing ever called PushRequest,
so the table was permanently empty and the handler was a parser and
nothing else.
Removed with the subsystem behind them: market.cpp, CProduct, CReview,
CUser, CReviewDB, the advert publish templates and the subscribe channel
machinery, none of which had an inbound handler either. MSG_REVIEW,
MSG_PRODUCT and MSG_TABLE stay in the inv enum, unserved: CInv puts type
on the wire as an integer and MSG_TX and MSG_BLOCK are shared with every
other node.
Unknown commands are already ignored for extensibility, so peers still
running these lose nothing but an answer they never asked for. Verified
by a clean build of bitflash-node.
(cherry picked from commit ad9c393866c007a115f32d70a9017ce91066a7d1)
|
|
std::thread::hardware_concurrency() reports the MACHINE's core count. Inside a
container that is the wrong number: with --cpuset-cpus 0-5 on a 12-core host it
still answers 12, and every caller that sized a pool from it oversubscribed 2x.
Measured on the pool box:
- the pool node logged "Mining with 11 thread(s) of 12 core(s)" while
confined to 6 CPUs
- a burst miner launched with /hashthreads=2 filled the ~2 GB RandomX
dataset with 12 threads, alongside the pool node in the same 8 GB container
AvailableCoreCount() asks sched_getaffinity what the cpuset actually permits,
which is the number all of these meant. It cannot make the dataset build slower:
the CPU budget is identical either way, so fewer threads on the same cores is
the same work without the thrash. Falls back to hardware_concurrency where the
call is unavailable, and never returns 0 because callers subtract from it.
Not the cause of the burst miner dying -- that was a corrupt Berkeley DB
environment in the worker datadir -- but it is why the failure landed on the
dataset build, and it is wrong on its own terms.
(cherry picked from commit b3bee16f3eef8982bff1174bd9d0a9c21473732c)
|
|
From the first of two wedges captured on the pool. It is a different
proximate fault from the second one, and this is the part that turns a
single bad socket into a node-wide outage.
A failed select() zeroes both descriptor sets, so no peer is serviced on
that pass. That is correct as far as it goes. What was missing is that
nothing stopped the offending descriptor coming straight back:
- the invalid socket was only marked fDisconnect, and a node is not
removed from vNodes until it has drained or DISCONNECT_DRAIN_SECS has
elapsed -- and it cannot drain, because its socket is gone;
- the select set was rebuilt from vNodes with no check, so the same dead
descriptor went back in on the very next pass and failed select()
again.
Measured in the capture: 1,027 consecutive 'its socket is no longer valid'
passes across 61 seconds -- the full drain timeout -- one bad socket, and
for the whole of it the node read nothing from anybody. The peer that
caused it had 57 bytes unsent, so it took the slow path out.
Now the descriptor is forgotten as soon as getsockopt says it is not ours:
select() succeeds again on the next pass and every other peer is served
normally while the dead node ages out as before.
Deliberately forgotten rather than closed. getsockopt has just said this
descriptor is no longer ours; calling close() on it is precisely how an
unrelated socket gets closed instead, which is the fault fixed in e742d0b.
(cherry picked from commit f14c513b4e39d6203b73584f6dc01acbb5a9546e)
|
|
The peer-height gate replaced a tip-age check that was wrong in both
directions. It is a strict improvement and it is still blind in the one
case that matters most.
When the node goes deaf -- peers still in vNodes, nothing arriving -- no
peer announces anything, GetPeerMedianHeight() returns -1, and the
fallback mines our best chain. That fallback is right for a node that has
not been told yet and wrong for a node that has stopped being able to
hear, and a height nobody sent cannot tell those apart. A wedged node is
exactly a node with no peer height information, so the gate was silent
for the whole of heights 3966-4085: 120 blocks, 5890.50 BTF voided.
Time can tell them apart. GetPeerHeightAgeSecs() measures how long since
any peer reported a height, counting from first use rather than the epoch
so a starting node does not refuse jobs before its first handshake. Past
20 minutes the pool stops issuing jobs and says so unconditionally in the
log. /poolmineblind overrides it: without an escape hatch a genuinely
isolated pool that is 90% of the hashrate could never restart its own
chain, which is the deadlock the tip-age check used to cause.
Also stop nStartingHeight from going stale. It was a snapshot taken once
during the version handshake and never touched again, so on a long-lived
connection the median is built from heights the peers had hours ago, and
the behind-the-network check drifts toward 'everyone is behind us' and
quietly stops firing. A peer that hands us a block is demonstrably at
least at that block's height.
(cherry picked from commit 0800c18faf703b4acde1b131d0b742d37a86d92b)
|
|
- RebuildJob: no jobs with zero peers or a tip older than 30min;
mining unsynced produced orphaned blocks and payouts that can
never be paid (observed in production: 8 orphan 'found blocks')
- payout exec: on SendMoney failure requeue with 60s backoff, up to
30 attempts, instead of silently forfeiting the payment
- nostr: add oxtr.dev, nos.social, fmt.wiz.biz, current.fyi, nostr.bg
(12 relays total) for announcement redundancy
(cherry picked from commit f1986cd7e4a6f69469a2256eb96ff03f3f146792)
|
|
RandomX reads its 256 MB cache, 2 GB dataset and per-VM scratchpads at random,
so most accesses miss the TLB on 4 KB pages. Requesting 2 MB pages recovers
roughly a tenth of the hash rate. Each allocation tries large pages first and
falls back to normal pages when the pool is not reserved, so nothing regresses
when hugepages are unavailable. -nolargepages opts out, and the diagnostics
report which allocations got large pages.
|
|
|
|
Review follow-up. A managed-Tor node receives every inbound onion peer
over its local hidden-service listener, so they all accept as 127.0.0.1.
Capping loopback would cap onion peers as one group and refuse the 33rd
legitimate onion peer -- the wrong limit for a network moving toward
onion-only. Exempt loopback (127.0.0.0/8): the global MAX_CONNECTIONS
still bounds the total, and the direct-IP flood this defends against
never arrives over loopback.
Bench: a 40-connection flood from 127.0.0.1 is now held in full (no
per-IP refusal), while a flood from the machine's LAN address still caps
at exactly 32.
|
|
A single IP could hold every inbound slot up to MAX_CONNECTIONS -- which
is what a connection-exhaustion flood did to the rendezvous relays, one
address parking hundreds of sockets and starving real peers. Add
MAX_CONNECTIONS_PER_IP (32): the accept path counts existing connections
from the same address and refuses a new one past the cap. The relays'
firewall connlimit does the same a layer lower; this makes every build
carry the defense, including future poll()-based relays with no FD ceiling.
|
|
|
|
|
|
|
|
|
|
|
|
# Conflicts:
# src/makefile.mingw
|
|
|
|
|
|
|
|
A registration was consumed by the first caller and registering again replaced the previous one, so a node was reachable to one peer at a time and unreachable for the gap between being paired and registering again. Measured on a live node: descriptors resolved 53/54, our own registrations paired 11/12, nothing paired and went silent -- and half of every outbound dial failed, 15 of 30 in eleven minutes.
The relay keeps a queue per pubkey capped at four, pairs with the oldest still-alive registration and discards the dead; the node runs three accept threads, each holding its own registration. The wire protocol does not change, so either side can be deployed first.
Bench with both relays built from source and waves of concurrent dialers: old+1 65%, new+1 58%, old+3 68%, new+3 98%. Each half alone lands within noise of doing nothing.
|
|
CNode::Disconnect() closed the socket and left the handle number in the object; ~CNode() then closed it again with a raw closesocket() the accounting never saw. There is no assignment of INVALID_SOCKET to CNode::hSocket anywhere in this tree, so the destructor's guard never fired. Between the two closes the OS can reissue that number, and the second close then takes down somebody else's connection. The same omission left a closed handle in the select() set, which fails every iteration and drives the recovery path that prints 'dropping node: its socket is no longer valid'. Measured A/B on the real network at matched socket volume: unpatched 552 sockets, 4 closes failing with 10038, 741 invalid-socket drops (all one peer), 11 peers held; patched 547 sockets, 0, 0, and 21 peers held. No protocol change.
|
|
A live node reported one close failing with 10038 (WSAENOTSOCK) at the tunnel's pump end. It leaks nothing; what it says is that a site's bookkeeping and its sockets disagree about ownership -- and Windows reissues a handle number only after its last close, so a site whose bookkeeping outlives its socket will eventually close a number that belongs to a live connection. BtfSocketTag now counts a handle arriving while another site still claims it, which names the site to go and read.
|
|
Windows lists a wildcard-bound socket twice in MSFT_NetTCPConnection, so counting Bound rows makes the number climb whenever the node talks to more peers -- 35 of a node's 38 Bound rows shared a local port with one of its own established connections. Counting orphans instead: a 1.2.10 node at 26h held 1262 and was still taking on 30-90 an hour; nodes on 1.2.11 and 1.2.12 held 3, transient. The leak was real and 1.2.11 ended it. Adds a live-unconn column (getpeername on every tagged socket, reports 1: the listener) and rewrites the notes that said otherwise.
|
|
The 1.2.11 accounting answered 'which site' and stopped. On a node with 3h51m of uptime it reported 58 live sockets while Windows attributed 112 to the process, 47 of them bound and never connected, against 55 closes that had returned an error -- with nothing recording what those errors were. Close failures are now counted per code per site, and the table totals, so the live figure can be compared directly against what the OS attributes to the process. Instrumentation only.
|
|
A node was found holding 1502 sockets in Windows bound state, one ephemeral port
each, 9% of the machine range in nineteen hours. Reading the code did not find
it: every connect() failure path here closes its socket.
So count. Every socket is tagged at creation and untagged at close, across all
ten sites, with a counter for closes that fail. Already rules out three
explanations: nothing is created or closed behind the counter (untagged closes
are zero), no close is failing, and the leaked sockets are not in the live set --
tracked live matches what the OS reports as established, while the bound pile is
something the program already closed successfully.
|
|
* Bump to 1.2.11
One change, and it is the reason a node on Windows kept collecting peers it
could not hear.
FD_SETSIZE was never defined in this tree, so it defaulted to 64 and select()
could watch the listen socket plus 63 peers -- everything past that stayed
connected and was never read again. Windows FD_SET stops adding once its array
is full and says nothing, so there was no error to notice. Because new
connections go on the end of vNodes, it was always the newest connection that
starved, including the ones dialed to fix connectivity.
Measured on two nodes: 277 and 90 CNode objects against 93 and 90 live
sockets, 316 and 355 sockets in a bound-but-never-connected state, and a status
bar reporting hundreds of peers for a node really talking to thirty.
1.2.11 raises FD_SETSIZE to 1024 on Windows, caps connections at 125, drops a
node stuck undeliverable after 60 seconds instead of holding it forever, stops
answering a failed select() with POSIX code that has no meaning on Windows, and
warns when there are more peers than select can watch.
* diag: let the node say what is wrong with it
Every hard problem in this project so far was found from outside the node --
sockets read over WMI, committed memory compared against a README, a grep over
somebody debug.log. In each case the node knew and had no way to say so.
One report, rendered identically for the GUI panel and the log: peers held vs
peers actually watched by select(), blocks received and how many arrived without
a parent, proof-of-work mode with the live miner thread count, and per peer the
age of the last message in each direction plus anything stuck unsent.
|
|
|
|
Two defects that feed each other. Measured on a 32-core production node:
1287 of 4117 processed blocks arrived with a missing parent, 97 arrived
that the node already had, and 490 transactions arrived whose inputs were
unknown.
mapOrphanBlocks had no ceiling at all. Orphan transactions have had one
since 0.1.0 -- MAX_ORPHAN_TRANSACTIONS with LimitOrphanTx -- and blocks
never did. An orphan block whose parent never turns up stays in memory for
the life of the process, each one holding a whole block, and a peer can
send an unlimited number of them just by inventing the previous-block hash.
LimitOrphanBlocks mirrors the transaction one: random eviction, both maps
kept in step, and the CBlock deleted because mapOrphanBlocks owns it.
Every orphan block also fired a getblocks at the sender, and an inv for a
block already held as an orphan fired another, with no filtering. So a node
missing part of the chain asked the same peer the same question over and
over, and the duplicate answers made more orphans, which asked again.
CNode::PushGetBlocks remembers the last request and drops a repeat -- the
same fix Bitcoin made in e2c2648c1.
One trap found by testing rather than by reading: eviction picks at random
and may pick the block just inserted, so GetOrphanRoot has to be read out
of pblock before LimitOrphanBlocks runs, not after. The first version had
that backwards and would have read freed memory.
Verified with fabricated orphans, four of the seven sharing one
hashPrevBlock so the multimap path is exercised: 7 in, limit(3) leaves 3,
limit(0) leaves 0, both maps in step at every stage, no crash. Then a run
against the live network to 4348 with no scaffolding left behind.
The storm itself was not reproducible on demand -- the network is healthy
today and produced one orphan block in seven minutes -- so there is no
before/after on the metric that motivated this.
|
|
A node running this code cannot tell "in sync with everyone" from "not
hearing anyone". The version message carries no height -- 0.1.0 never had
the field and this fork inherited the gap -- so the only evidence of a
problem is a number on screen that stopped moving, which looks the same as
a network with nothing to report. It is the structural reason a broken node
here has always been indistinguishable from an idle one.
Append nBestHeight to the version message and keep what a peer announced in
CNode::nStartingHeight. The field goes last, so a node that predates it
stops reading after addr and treats the rest as trailing bytes, which cost
it one log line. Reading it is guarded by vRecv.empty() and never required:
peers that do not send it keep nStartingHeight at -1 and are left out of
the median. Every release so far sends the short form.
GetPeerMedianHeight() takes the median rather than the maximum so a single
peer claiming an absurd height cannot move it.
Then use it. After five minutes more than one block behind the peer median,
the node says so in the log -- not behind a debug category, because an
operator who already enabled -debug is not the one who needs telling -- and
the GUI carries the same sentence in its status area. "8 blocks behind for
27 minutes" is something a user can act on; a frozen counter is not.
Verified against two nodes on the live network: the one starting from an
empty datadir read its peer at 3797, reported 2636 then 1274 then 433
blocks behind as it caught up, and went quiet on its own once level. The
other logged "peer height = -1" for the older nodes it met on the real
network, which is the compatibility path doing its job.
Known limit: nStartingHeight is a snapshot from handshake time and goes
stale on a long-lived connection, so this catches a node that is receiving
something but not blocks, not one that has gone completely deaf. The
inactivity timeout from #55 covers that case and refreshes these heights
when it forces the reconnect.
|
|
(#55)
A connection whose path dies silently -- NAT drops an idle mapping, a relay
restarts, a route changes -- never delivers FIN. Nothing becomes readable,
no error is raised, and the node simply stops hearing from that peer while
the socket looks perfectly healthy. #51 bounded this for the rendezvous
registration; the connection that comes out of the rendezvous had the same
hole, and so did every ordinary peer.
Stamp nLastRecv and nLastSend on the socket path and act on them:
- nothing at all within 60s of connecting, disconnect. In testing this
fired twice per run against dead .btf descriptors handed out by the
relay, which is where the ghosts have been coming from.
- queued to send for 10 minutes with nothing going out, disconnect.
- nothing received for 30 minutes, disconnect. A healthy peer relays a
block inventory about every two minutes, so this is fifteen intervals
of silence. Bitcoin used ninety minutes against ten-minute blocks.
Disconnecting is the whole cure: reconnection already works, and a peer we
cannot hear is worth what no peer is worth.
A node with nothing to relay would now look dead to a peer applying the
same rule, so send "ping" after ten minutes of having said nothing. It
carries no payload and expects no answer -- arriving is the entire content,
because the receive path has already stamped nLastRecv by the time it is
parsed. Requiring an answer would have read every node predating this
change as dead, since those ignore unknown commands rather than replying.
Verified on the live network: synced 0 to 3790 with five peers and no
spurious disconnects, then again with the interval cut to 20s, which put 54
pings on the wire with no peer dropping us and no exceptions.
|
|
Satoshi left a note above ThreadBitcoinMiner in 2009:
//// todo: start one thread per processor,
//// use getenv("NUMBER_OF_PROCESSORS")
Nobody did, and it stopped being a nicety when the proof of work
became RandomX. SHA-256d on a 2009 desktop lost little to a
single thread. RandomX is built to be fed by every core at once,
so one thread on a 32-core machine mines at a thirty-second of
what the box can do -- and the only outward sign is a CPU that
looks suspiciously idle. A user reported exactly that: 30 minutes
of "Mining", no blocks, Bitflash sitting at 3.3% CPU while the
2 GB dataset was allocated and fast mode was active. Nothing was
broken. It was hashing on one core.
bitflash -gen # every core but one
bitflash -gen -genproclimit=8 # exactly eight
Automatic leaves one core to the rest of the node -- accepting
blocks, talking to peers, drawing the window -- because a node
that mines beautifully while falling behind the chain has won
nothing. /genproclimit overrides it, and a count several times
the core count is refused rather than quietly thrashing.
The wrapper was already built for this: RandomXCreateMinerVM says
"own VM, not shared", and the ~2 GB dataset is one allocation
every thread reads. Only the 2 MB scratchpad multiplies, so
sixteen threads cost about 32 MB more than one. The startup line
says so, because "why is it using 2 GB" is the first question
anyone asks.
Two things had to be right for this to be safe:
- Shutdown. Miners share slot 3 of vfThreadRunning and StopNode
waits for every slot before DBFlush tears the database down.
With several, the first one out announced that everyone had
finished while the rest were still calling ProcessBlock, so
they are counted now and only the last one clears the flag.
The three CheckForShutdown(3) calls inside the miner became
plain returns: _endthread() does not unwind on Windows, so
the bookkeeping would have been skipped there.
- Duplicated work. Each thread makes its own key, and the
coinbase pays it, so every thread has a different merkle root
and searches a different space. Block assembly was already
under cs_main and cs_mapTransactions; AddKey and ProcessBlock
take their own locks.
Verified on both platforms: four distinct thread ids reach the
hashing loop and all report back on shutdown, which exits 0 with
DBFlush running. Auto picks cores-1 (1 on a 2-core box, correctly
reported). Speedup itself needs a many-core machine to show and
has not been measured here -- the claim in this message is about
threads doing work, not a benchmark.
Not touched: RandomX's memory. RANDOMX_DATASET_BASE_SIZE and
SCRATCHPAD_L3 are part of the hash function, so a knob for "use
more RAM" would either do nothing or fork the chain. Threads are
the only real dial.
|
|
A node registers at a relay and waits to be dialled. The wait
had no bound, and ConnectTo() clears the socket timeout on
purpose -- a registration is idle by design and may sit for a
long time.
What that cannot survive is the path dying with neither end
saying so, which is the ordinary case and not the exotic one:
a NAT drops the idle mapping, a relay restarts, a route
changes. No FIN arrives, nothing becomes readable, and recv()
never returns. The thread stays alive and asleep, so
ThreadBtfAccept never reaches the `continue` that would
register again. The node is unreachable and every outward sign
says it is fine.
Caught on the bootstrap seed: 27 minutes without a block,
eight behind the network, relay reachable the whole time, and
a restart fixing it instantly. Nothing in the log marked the
moment it went deaf.
I tried TCP keepalive first and it does not work here. With
probes every 5s and a 2-probe limit, a socket whose replies
were being dropped was still ESTAB after four minutes with the
probe counter stuck at zero. It is measured, not assumed, and
that approach is not in this change.
So the bound is ours to enforce, at the one place that knows
what waiting means. RvServiceWaitPaired takes a timeout;
expiring is not an error but "nobody came, go round again",
which the loop above already handles. Cost is one reconnect
per relay per five minutes.
Verified by killing a registration's path silently -- dropping
only the relay's replies, so no RST or FIN is ever seen. With
a 60s bound for the test:
rendezvous: no dial at 92.246.128.180:8434 within 60s
(or it dropped us), re-registering
Registrations went 1 -> 2 and the node carried on. Before this,
the same treatment left it blocked past four minutes with no
log line at all.
The socket goes back to blocking once a dial lands, because
from that point it is a data tunnel and a tunnel is allowed to
sit quiet between messages.
|
|
The segfault in issue #5 was a format string: `%I64d` is MSVC-only,
glibc parses it as the GNU `I` flag with a width of 64 and an `int`
conversion, the vararg run desynchronises, and the following `%s`
dereferences an integer. It killed the node mid-sync at the first
block carrying a real transaction, deterministically.
GCC can catch this. It never got the chance, for two reasons that
had to be fixed together:
- `-w` sat in both makefiles and suppressed every diagnostic. It
cannot be narrowed; it wins over any -W or -Werror= that follows,
so it had to go. The four -Wno- entries replacing it are the
categories this 2009-era tree actually produces (measured: 248
deprecated-declarations, 157 write-strings, 18 narrowing, 1
unused-result), silenced by name instead of wholesale.
- Even with -Wformat on, nothing was checked. GCC only inspects
format strings for functions it knows are printf-like, and every
format string here goes through strprintf, error, my_snprintf, or
OutputDebugStringF -- `printf` itself is #defined to the last of
those. None carried a format attribute, so the entire tree was
invisible to the check. They are annotated now.
Annotating surfaced 77 warnings across 17 distinct sites, all fixed
here. Most were size_t or int64 passed to %d, harmless on x86-64 but
wrong. Two were not cosmetic:
- CBlockIndex::ToString printed pprev/pnext with %08x, truncating
both pointers to 32 bits, so the two fields that line exists to
show were unreliable on every 64-bit build.
- CInv::GetCommand threw "type=% unknown type", where the % ate the
space as a flag and the u as its conversion. The message read
"type= 3nknown type".
PRId64 is now one spelling for every target. Its MSVCRT branch was
what produced "I64d" in the first place, it is pointless against UCRT
which takes the C99 forms, the rest of the tree already prints int64
with %lld on Windows, and its only user (i64tostr) is dead code.
Collapsing it lets both builds share the gnu_printf archetype --
ms_printf would reject %zu, which this tree uses and UCRT supports,
while waving through the %I64d that crashes.
Verified on both platforms: clean build, and reintroducing the
issue #5 specifier now fails compilation rather than shipping.
main.h:410:50: error: unknown conversion type character 'I' in format
main.h:410:70: error: format '%s' expects argument of type 'char*',
but argument 3 has type 'long long int'
That second line is the crash itself, caught at build time.
Runtime checked on both: node syncs past the block from issue #5, and
the repaired log lines read correctly --
before: CTxOut(nValue= 50.%08I64d, scriptPubKey=(null))
after: CTxOut(nValue=50.00000000, scriptPubKey=0xE1C7EF3E706002)
Four '_WIN32_WINNT redefined' warnings remain on Windows. Silencing
them means letting a system header pick the Windows API level instead
of this tree, which is a separate decision from format checking.
|
|
* net: compiled-in bootstrap seeds, so a first run does not need a relay
A node with no cache has only the Nostr relays. If they are down, blocked or
slow, a first run has no way into the network at all -- and that is the one
moment a user has no patience for. Seeds are the floor under that.
A seed entry is an address and an encryption key. Deliberately no meeting node:
the rendezvous relay pairs on the service's public key, and a `.btf` address is
that key, so a client can find a seed by trying the known relays in turn.
Recording which relay a seed was on would rot the moment it failed over, which
is exactly the failure this project spent weeks chasing -- advertising a
rendezvous somebody is not registered at.
Seeds carry no authority. They serve the same signed, self-certifying
descriptors as any peer, so a hostile seed can stall a bootstrap but cannot
forge a peer or feed a false chain.
The compiled list is empty. Which nodes become permanent public bootstrap
targets is a decision for whoever runs the network, not a default, and it is
close to irreversible once binaries are distributed. Empty is valid: discovery
falls back to Nostr exactly as before. /btfseed=ADDRESS:ENCHEX (repeatable)
appends at runtime, for testing and for private networks.
Tested against the live network, clean datadir, no cache, three seeds carrying
no meeting node:
btfpeers: no cache yet, falling back to seeds
btfseed: trying 3 seed(s) across 3 relay(s)
btfseed: reached q5jxyzn...btf via 90.156.222.107:8434
btfseed: 3 seed(s) answered
* net: add the Almaty bootstrap node to the seed list
Dedicated node, no wallet balance, no mining, running beside the rendezvous
relay on the same host. It exists only to answer a first dial.
Fresh datadir, no flags, compiled list only:
btfpeers: no cache yet, falling back to seeds
btfseed: trying 1 seed(s) across 3 relay(s)
btfseed: reached ygffz37...btf via 90.156.222.107:8434
btfseed: 1 seed(s) answered
btfpeers: kept 2 of 2 descriptor(s)
5.2 seconds from nothing to a peer, against 36s for a cold start that had to go
through the relays. The seed then handed over two more peers by exchange, which
the cache keeps for next time.
|
|
Discovery has depended entirely on the Nostr relays. Two connected nodes never
told each other who else existed, so a relay outage left a running network
unable to grow, and a node that found one stale peer had no way to learn there
was anything better. The old `addr` gossip went out with the IRC seed.
Peers now trade signed descriptors over a new "btfpeers" message: our own
first, then peers that actually answered us. The receiver checks each Schnorr
signature against the key the `.btf` address decodes to, so the sender is
trusted for nothing -- forging an entry needs somebody else's secret key. That
leaves flooding with self-generated keys, which the per-message caps and the
"peers that answered" ordering are there to blunt.
Unknown commands are ignored (main.cpp), so nodes without this just drop the
message; nothing here is a protocol break.
The cache from the peer-cache change now keeps each peer's descriptor verbatim
alongside the parsed fields, because that signature is what makes a peer
relayable. Entries without one -- resolved through Nostr, or written before
this -- stay dialable, just not passed on. A dial never clears a descriptor
already on file.
|
|
A node could not become reachable. RvServiceRegister connected, sent its
'S'+pubkey header, and then blocked reading the relay's 0x01 byte -- which the
relay only writes when a client pairs with the service. So the call did not
return when the node was registered; it returned when somebody had already
dialled it.
ThreadBtfAccept called BtfSetActiveRelay after that. The result is a deadlock:
- to be dialled, a node must publish a descriptor naming its meeting node
- to publish, BtfActiveRelay() must be set
- it is only set once somebody has dialled
Before #26 this was masked. BtfActiveRelay() fell back to
vBtfMeetingRelays[0], so a node always published something, and ThreadBtfAccept
picks its relay at random -- so it worked whenever the two happened to match,
roughly one time in the number of relays. That is the flakiness this project has
been chasing: not a bad relay, an advertisement that usually named the wrong one.
#26 removed the fallback, which was right on its own terms and turned an
unreliable heuristic into a reliable deadlock. This is the other half of that
change and they belong together.
RvServiceRegister now returns as soon as the relay has the service listed.
RvServiceWaitPaired does the blocking read. ThreadBtfAccept registers, advertises,
then waits. RvClientConnect is untouched -- a dialer blocking until it is paired
is correct.
Registration failure also had no log line at all, which is why this stayed
invisible: a node that never registered looked identical to one waiting quietly
for a peer. There are now lines for registering, failing over, and being dropped.
Verified: a fresh node reports `registered at 90.156.222.107:8434, waiting for a
dial` and then self-resolves with `meeting_node=90.156.222.107:8434` -- the relay
it is actually on. Two nodes on one machine reach each other.
|
|
PR #25 was opened, reviewed and approved as a single-file change to net.cpp:
the .btf peer cache, +169/-0. I then pushed peer exchange and a /port option
onto that same branch while the PR was open, which silently widened it to six
files. The merge carried all of it in.
Nothing about the peer exchange had been validated. Its two-node test never
completed -- node B could not dial node A at all -- so what landed on main is
unproven network protocol code that no one agreed to take.
This restores main to what #25 actually described. The peer cache stays, with
its runtime numbers (36.3s to 2.1s to first peer). #26 stays. What comes out:
- the "btfpeers" message, both directions
- BtfPexCollect / BtfPexAccept and the per-message caps
- CNode::nLastPexRecv
- the `desc` field on the cached peer record
- BtfLocalDescriptor / BtfSecpContext
- /port
Peer exchange returns as its own pull request once its test passes. The /port
option is worth having on its own -- two nodes cannot share a machine without
it, which is what blocked the test -- and should go up separately rather than
ride along again.
Release v1.2.2 is unaffected: it was built from ba2de82, before either merge.
|
|
* net: remember .btf peers that answered, and dial them first on start
Discovery depends entirely on the Nostr relays. A node that has been up
for hours knows exactly which peers respond, and throws all of it away on
exit; the next start reaches for the relays again and walks the same
descriptor list, of which roughly 96% is dead. That is what makes a
restart take minutes before the first tunnel opens -- measured today at
about seven.
Every peer that completes a tunnel is now written to btfpeers.json with
its meeting node and encryption key, which is everything needed to redial
it. On start a thread tries those before the relays have answered
anything, stopping once eight are connected. Entries older than a week
are dropped, and the file holds at most fifty.
Cache only: nothing here changes how a first-ever run bootstraps, and a
missing or corrupt file just falls through to relay discovery.
* net: exchange signed .btf descriptors between peers
Discovery has depended entirely on the Nostr relays. Two connected nodes never
told each other who else existed, so a relay outage left a running network
unable to grow, and a node that found one stale peer had no way to learn there
was anything better. The old `addr` gossip went out with the IRC seed.
Peers now trade signed descriptors over a new "btfpeers" message: our own
first, then peers that actually answered us. The receiver checks each Schnorr
signature against the key the `.btf` address decodes to, so the sender is
trusted for nothing -- forging an entry needs somebody else's secret key. That
leaves flooding with self-generated keys, which the per-message caps and the
"peers that answered" ordering are there to blunt.
Unknown commands are ignored (main.cpp), so nodes without this just drop the
message; nothing here is a protocol break.
The cache from the peer-cache change now keeps each peer's descriptor verbatim
alongside the parsed fields, because that signature is what makes a peer
relayable. Entries without one -- resolved through Nostr, or written before
this -- stay dialable, just not passed on. A dial never clears a descriptor
already on file.
* main_gui: implement /port, which net.cpp already documented
net.cpp has said nListenPort is "tunable via /port" since it was written, but
nothing ever parsed the option. Two nodes could not share a machine, which is
exactly what testing peer exchange needs. Sets addrLocalHost.port too, so we
do not advertise a port we never bound.
|
|
* Add Linux GUI installer via Wine
* Fix launcher creation and add BTF_RELAY env var support
* Pool server, Linux GUI, unified binary, README update
* Root Makefile: make linux / make windows, cleaner README
* Remove headless.cpp
* Fix XPM color count mismatch causing toolbar image errors on Linux
* XPM icons: same code on Windows and Linux, no ifdefs around image loading
* Fix GTK icon crash and Linux printf format
* btfrv: 10s socket timeout on relay connects
* Faster peer discovery: more dials per pass, shorter backoff and sleep
* Parallel btf peer connections
* Fix GetMyExternalIP - was disabled, now uses modern IP echo services
* Fix tunnel disconnect on idle and restore external IP detection
* Filter stale Nostr descriptors
* Sort peers by freshness, fresh first
* Pool: add .btf Stratum listener — miners connect via rendezvous, no port forwarding needed
* btf pool: self-replacing listeners, one per relay, no queuing between miners
* Mining modes: solo/operator/participant — pool server on-demand, participant Stratum client
* Status bar: show external IP and .btf address
* README: clean rewrite, concept-first, minimal
* make linux produces AppImage
* make linux produces AppImage
* Pool dashboard: connect instructions, tabs, share bars
* Perf fixes: GetMyExternalIP threaded, btf listener counted
* Perf fixes: GetMyExternalIP threaded, btf listener counted
* Fix pool payouts: defer until coinbase maturity
* Pool: clean mode switching, IP warning
* Pool: clean mode switching
* Fix gPoolRunning: remove static
* AppImage: use real icon from bitcoin.ico
* Fix exit crash; Mining menu
* Fix exit crash; Mining menu
* Fix exit crash; Mining menu
* Fix exit crash; Mining menu
* Fix exit crash; Mining menu
* Windows build: use mingw32-make
* Windows build: simpler make invocation
* Add deps-windows target
* Windows build fixes
* Windows build fixes
* Windows makefile: fix obj dir
* Windows: output to repo root
* Windows: bundle DLLs into zip
* Replace wxWidgets with ImGui+GLFW: same build on Linux and Windows
* Fix %I64d that slipped back through rebase (main.h, net.h)
* MinGW: add _mkdir shim for non-MSVC Windows; fix %I64d
* Fix gPoolRunning extern outside ifndef _WIN32
* Remove -lboost_system: boost is header-only in this codebase
* Status bar: copy buttons for IP and .btf address
* Windows: add app icon via windres, strip wx stuff from rc file
* Clean GUI: single Options menu item, bigger font, better spacing, cleaner options dialog; update README
* GUI: embed Roboto font at 16px for clear readability on all platforms
* Windows: windres depends on obj dir target
* Fix balance hidden off-screen: put on own line below address bar
* Windows: inline mkdir in every rule, remove dir target that confused make
* Windows: move all target to top so it's the default rule
* Fix pool server not starting: always start when operator selected, re-sync dialog state on open
* Write debug.log to data directory (AppData/Roaming/Bitflash on Windows)
* Participant miner: use wallet address as username so payouts go to wallet
* Add .gitattributes to lock LF line endings, prevent CRLF drift
* Refactor pool mining to btf-only transport
* Persist pool settings and startup wiring
* Prune stale peers and expose pool queries
* Fix build regressions and add CLI mode/arg parity
* Fix rpc miner stats declaration order
* Centralize pool runtime extern declarations
* Separate operator/participant dashboards and restore peer stats
* Add live telemetry strip and clearer per-mode runtime metrics
* Reduce mining dashboard clutter and remove manual pool entry
* Count only authorized pool miners
* Seed initial pool job on startup
* Use a real share target for pool work
* Switch pool accounting to round-based shares
* Fix nostr pool announcer stats call sites
* Fix pool stats definition order
* Improve pool worker status display
* Add mutex include for participant status
* Replace ImClamp in dashboard layout
* Clarify participant work status
* Log participant status transitions
* Reduce peer discovery log noise
* Add participant status badge
* Track participant auth acknowledgements
* Log and validate pool job delivery
* Stop persisting mining mode/pool settings across restarts
LoadWallet() runs after CLI flags are parsed in main_gui.cpp, and was
restoring nMineMode/strParticipantPool/strPoolName/strPoolDashboardUrl/
dPoolFeePercent/fGenerateBitcoins from wallet.dat every time, silently
overwriting whatever was just set on the command line or in the current
Options session -- no error, no log line. This is why pool name/address
could show correctly in the UI but stay stale in the logs (UI held the
in-memory value just set; logs eventually reflected whatever LoadWallet
clobbered it back to), and why mode changed behavior across restarts.
Mode/pool config is now decided fresh every launch: CLI flags if given,
otherwise the compiled-in default. Nothing is written back to wallet.dat
for these settings anymore, so there's nothing left to go stale or fight
over load order.
* Add explicit Relay mode as the new startup default
4th mining mode: node runs/syncs, never mines, no pool server -- an
explicit option instead of an implicit side-effect of 'Solo + never
click Start Mining'. Now the compiled-in default (was Solo).
- main.h: MINE_RELAY = 0, renumbered other modes (nothing compares
nMineMode against raw integers elsewhere, verified).
- main.cpp: default nMineMode = MINE_RELAY; BitcoinMiner() early-returns
for Relay, mirroring the existing Participant-mode guard.
- gui.cpp: added Relay radio button, status label, and telemetry-strip
text. Start/Stop Mining button is hidden for Relay (meaningless -- it
never mines by definition); OK forces fGenerateBitcoins off if you
switch into Relay while previously mining.
* Announce pool changes immediately instead of waiting up to 60s
ThreadBtfPoolAnnouncer re-read strPoolName/dPoolFeePercent/etc live, but
only once per iteration of its 60s sleep loop, so a pool name/fee change
in Options could take up to a minute to reach other nodes.
- nostr.h/.cpp: add gAnnounceNow flag; the announcer's sleep loop now
wakes early and publishes immediately when it's set, then clears it.
- gui.cpp: Options OK handler sets gAnnounceNow when operator pool
settings actually changed (name/dashboard/fee/mode), so a saved
change is announced right away instead of on the next tick.
* Give each pool miner a distinct extranonce2 so jobs no longer overlap
mining.notify previously broadcast one identical 80-byte header (baked
from a single server-side extraNonce) to every connected miner, so any
two miners on the same job searched the exact same nNonce space instead
of disjoint ranges.
Each Miner connection is now assigned a unique extranonce2 on connect.
BuildMinerBlock() takes the shared job template and appends that miner's
extranonce2 to the coinbase scriptSig, recomputing the merkle root, so
every miner gets its own header variant of the same job. mining.submit
validates and, on a full block, assembles the block using that same
miner-specific variant (via m->extranonce2), not the shared template.
No client-side protocol change needed: main.cpp's participant miner
already just takes whatever header bytes mining.notify hands it and
submits (jobId, nonce) back, so it works unmodified against per-miner
headers.
Still simplified vs standard Stratum (no coinb1/coinb2/merkle-branch
wire split, no mining.extranonce.subscribe reply payload) since this
protocol was already a custom simplification pre-existing this fix.
Syntax-checked with g++ -fsyntax-only (full link needs secp256k1/
RandomX/GLFW not present in this sandbox).
* Fix debug.log never being written on the real Windows build
printf-to-debug.log routing was gated on:
#if defined(_WINDOWS) || !defined(_WIN32)
_WINDOWS is only ever defined by the legacy, unused MSVC makefile.vc.
The actual Windows build (makefile.mingw) never defines it, and _WIN32
IS defined there (mingw builtin), so on real Windows this condition was
always false || false = false. printf() fell through to plain libc
printf (stdout only), so debug.log was never created, and output was
only visible if the process happened to be run attached to a console.
OutputDebugStringF() was already platform-safe internally (always
writes debug.log; only echoes to console when not on Windows), so the
outer condition was pure dead weight. Made the routing unconditional.
* Reduce base UI font size one more step (16px -> 14px)
* Remove Start/Stop Mining button; mining state follows selected mode
The button let a user configure a pool then click a separate toggle,
creating states where the button's label/action didn't match what the
selected mode actually needed. OK now starts/stops the miner thread
itself based on nMineMode: Relay clears fGenerateBitcoins, the other
three modes set it and spawn ThreadBitcoinMiner if it isn't already
running (guarded by vfThreadRunning[3]).
* Fix participant miner stuck waiting for notify after buffered multi-line recv
Root cause of the operator/participant pairing successfully but no work
ever flowing: PoolParticipantMiner()'s receive loop only pulled one line
out of the socket per select() wakeup. The server sends the authorize
reply, mining.set_difficulty, and mining.notify back-to-back on the same
connection, and these often land in a single TCP segment — one recv()
inside StratumRecvLine() buffers all three lines into 'buf', but only the
first was extracted and processed. The remaining lines (frequently
mining.notify itself) sat in 'buf' indefinitely, because select() only
reports readability of the OS socket, not whether our own buffer still
holds unprocessed complete lines — and the server had nothing further to
send once the initial burst was over, so select() never fired again.
Fix: drain any complete line already sitting in 'buf' first; only fall
back to select()+recv() when 'buf' has no complete line left.
* Fix mining.submit rejecting every share due to wrong param-count check
Server required params.size() >= 4 for mining.submit, but the client
only ever sends 3: [username, jobId, nonceHex] (indices 0-2, matching
what the handler actually reads: p[1]=jobId, p[2]=nonceHex). Every
submission was bounced with 'bad params' before its nonce was ever
checked against the job, producing an endless share-sent/share-rejected
loop with no other visible cause. Likely a leftover from the standard
5-field Stratum mining.submit shape, which this simplified 3-field
protocol never matched. Threshold corrected to size() >= 3.
* Fix mining.submit rejecting legitimate shares that weren't also winning blocks
Pool share difficulty is intentionally ~256x easier than full network
difficulty (MakeShareTarget() shifts the block target left 8 bits), so
miners are expected to submit many shares per actual block found. But
the handler replied to the miner based on ProcessBlock()/CheckBlock()
success — validation against the much harder network target — rather
than on whether the share met pool difficulty. Every legitimate share
that wasn't also a winning block (the vast majority, by design) failed
CheckBlock()'s nBits check and got reported to the miner as rejected.
Now: reply(true) as soon as a submission clears pool-share difficulty.
Only if it ALSO clears full network difficulty (job.target, not
job.shareTarget) does it get submitted as a block — with no second
reply sent, since the share was already acknowledged.
* Fix dead-peer dial priority: freshness sort was undone by full shuffle
ConnectDiscoveredBtfPeers sorted candidates newest-first, then immediately
shuffled the whole list, making the freshness sort a no-op — every pass was
fully random. Meanwhile g_btfPeerFails was tracked but never used to
prioritize, so a peer that had failed 20 times in a row got equal footing
with an untried or previously-successful peer; the only effect was backoff,
capped at 15 minutes, so dead descriptors cycled back into the dial pool
constantly.
- Track successful connections (g_btfPeerEverConnected) and use it, plus
fail history, to order candidates in tiers: known-good, then untried,
then known-bad (worst offenders last). Freshness order is preserved
within each tier instead of being discarded.
- Raise the backoff cap for chronic failures (6+ in a row) from 15 minutes
to 6 hours, so descriptors that are clearly dead stop consuming dial
budget every pass while still getting retried eventually.
* Batch-resolve .btf descriptors to stop hammering public relays during dial passes
Root cause of 'stuck at 1 peer' despite good peer discovery: each parallel
dial thread in ConnectDiscoveredBtfPeers called ConnectNodeBtf, which itself
walked all configured relays (relay.damus.io, nos.lol, relay.nostr.band,
relay.snort.social) to resolve that one peer's descriptor. With up to
BTF_DIALS_PER_PASS (12) threads dialing at once, a single pass could fire up
to 12 * nNostrRelays near-simultaneous fresh connections at the same handful
of public relays -- easily enough to trip rate limits or get handshakes
refused (exactly what the logs showed: 'handshake refused by relay.damus.io',
repeated connect failures to relay.nostr.band / relay.snort.social). Peers
that were genuinely reachable looked dead simply because their descriptor
couldn't be fetched under that load, and then got penalized as known-bad by
the dial-priority tiering from the previous fix.
- Add BtfResolveMany: resolves every still-unresolved candidate's descriptor
in a single Nostr REQ per relay (multiple authors in one filter), instead
of one relay connection per candidate. A 12-candidate pass now costs at
most a few relay round-trips total instead of dozens.
- Add ConnectNodeBtfResolved / ConnectNodeBtfTail (net.cpp): given an
already-resolved descriptor, dial straight to the peer's rendezvous node
with zero further relay contact. ConnectDiscoveredBtfPeers now batch-
resolves all candidates up front, then fans out tunnel-only connect
threads in parallel -- no thundering herd against the relays.
- Candidates that fail to resolve are recorded as a failure immediately
(same backoff/fail-count bookkeeping as before) instead of spawning a
thread with nothing to dial.
* Stop reconnecting to relays constantly: pool one persistent connection per relay
Real root cause of the ongoing 'handshake refused' / 'connect failed' spam
and peer count stuck at 0-1: every relay-touching operation (startup self-
test, per-pass seeding, batch descriptor resolve, pool announce, pool query)
opened its own brand-new CWebSocket every single time it ran. Two independent
threads (ThreadNostrSeed and ThreadBtfPoolAnnouncer) each reconnect from
scratch on their own ~60s cycle, so a node could fire well over a dozen
fresh TCP+TLS+WS handshakes at the same 4 public relays inside any given
minute. Public relays -- relay.damus.io especially -- rate-limit new
connections per source IP fairly aggressively, so this node was effectively
DoS-ing itself: peers were genuinely reachable, but their descriptors
couldn't be fetched because the relay had already started refusing this
node's own repeated reconnects.
- Add a small per-relay connection pool (g_relayPool) and an RAII
CRelayLease that reuses a relay's live socket across calls instead of
reconnecting every time, only reconnecting when a socket actually drops
or was never established.
- Route every relay call site through it: the startup self-test, the
self-test loop, SeedFromRelay, BtfResolve, BtfResolveMany,
BtfPublishPoolAnnouncement, BtfQueryPoolAnnouncement.
- Only log 'connected to <relay>' on an actual fresh (re)connect, not on
every pass that's just reusing an already-live socket, so the log
reflects what's really happening instead of implying constant churn.
This cuts fresh relay connections from several per minute down to
essentially one per relay for the life of the process (until a relay
actually drops the connection), which should make the rate-limit-driven
refusals go away and let descriptor resolution -- and therefore dialing --
actually succeed at the rate the peer list implies it should.
* Fix wallet crash on rejected send; remove legacy IP-based peer discovery
- SendMoney() no longer throws when a committed tx is rejected by
AcceptTransaction() (e.g. raced input already spent). It now unwinds
the wallet state via RollbackTransactionSpent() and returns false,
which the GUI already handles as 'Transaction failed.'
- Removed the dead IP-based peer discovery/connect path now that
discovery is entirely Nostr/.btf: CAddrDB/addr.dat, mapAddresses,
AddAddress, ConnectNode(CAddress), FindNode(CAddress), ConnectSocket
(SOCKS4 proxy helper), addrProxy, ThreadOpenConnections/2 class-C
dialer and its thread spawn, the addr/getaddr P2P handlers, and the
dead vAddrToSend/setAddrKnown fields on CNode.
- Deleted irc.cpp/irc.h (already-disabled legacy seed code) and
scrubbed references from headers_core.h and all four makefiles.
* Raise peer connection target and add dial diversity; fix IPv4-only relay connect
- BTF_TARGET_CONN: 8 -> 24. With ~100 known .btf peers, average degree
needs only to exceed ln(N)~=4.6 for the network to stay connected
with overwhelming probability (Erdos-Renyi); 24 is a healthy margin.
- ConnectDiscoveredBtfPeers now splits each pass's dial budget ~70/30
between the existing reliability-tiered order and a random sample
from the full fresh peer pool. Pure reliability bias means every
node converges on the same 'known good' hub set, which correlates
peer sets across the network -- a relay outage then drops the same
edges everywhere at once instead of independent ones. The random
slice keeps a floor of uncorrelated edges.
- CWebSocket::Connect (relay websocket) now resolves with AF_UNSPEC
and tries every returned address (IPv6 or IPv4) in turn instead of
hard-failing on AF_INET-only resolution, which was one plausible
cause of the relay.nostr.band:443 connect failures (CDN-fronted
relays increasingly rate-limit/filter the v4 path).
* Remove accidentally committed build binary; ignore src/bitflash
* Fix pool participants getting stuck after the first job
JobWatcherFn only rebuilt/broadcast a new Stratum job when a new block
arrived on this node or no job existed yet. Once a job was built after
startup, gHaveJob stayed true forever unless *this* node found the
block, so with no new block (slow interval, or this node just not
seeing one) every connected miner sat hashing the exact same job --
stale timestamp, stale mempool snapshot -- indefinitely. From the
miner's side that's indistinguishable from being stuck: it worked at
connect (gets the one-time job sent on authorize) and then nothing
ever arrives again.
Added a 60s periodic refresh so the job (and therefore mining.notify)
gets rebroadcast on a timer too, same as a normal stratum pool, not
only on new-block/no-job events.
* Diversify Nostr discovery relay list
Went from 4 relays to 7, and reduced Cloudflare concentration: previously
2 of 4 (damus.io, snort.social) sat behind the same CDN, so a Cloudflare-
side issue could take out half the list even though they're nominally
independent relays. Kept damus.io (most established) as the one CF-fronted
entry and filled the rest with different hosts (nos.lol, nostr.band on
Hetzner, nostr21.com, offchain.pub, nostr.mom, primal.net) so peer
discovery doesn't have a single shared upstream as a point of failure.
7 relays is within the commonly recommended 5-10 range for redundancy.
* Fix data race in RebuildJob: chain-state reads weren't holding cs_main
pindexBest, nBestHeight, GetNextWorkRequired() and GetMedianTimePast()
were all read without cs_main. A block being accepted concurrently by
the P2P thread could interleave with this: pindexPrev captured from the
old best block, but nBestHeight already bumped to the new one by the
time job.height is computed, or pindexBest reassigned mid-read. That
mismatch would make QueuePayouts() schedule a payout for the wrong
maturity height for whatever block gets found against that job.
Wrapped the whole chain-state snapshot in one CRITICAL_BLOCK(cs_main),
consistent with how the rest of the codebase touches this state.
* add operator fee
* Add categorized, burst-collapsing logging for pool/worker/payout flow
debug.log was growing unboundedly because every printf() call wrote
unconditionally, so loops that log once per item (discovered peers,
connect attempts) produced a near-identical line per address with no
cap. Add LogPrint(category, ...): a drop-in printf replacement that
tags each call with a subsystem (pool/worker/payout/net/nostr) and
collapses repeated calls from the same call site into one line per
3s window plus a '(xN similar in 3s)' summary, so log growth is
bounded by elapsed time instead of event count.
Wire it through the whole pool -> work -> payout pipeline (rpc.cpp
operator side, main.cpp participant/worker side, net.cpp/nostr.cpp
peer & relay discovery), and along the way fix several previously
*silent* failure points that produced exactly the reported symptom
(worker stuck asking for work, nothing logged):
- Relay listener registration failing: no log at all, so a pool with
every relay down looked identical to a healthy idle pool.
- No relays configured: silent infinite 5s retry loop.
- Initial job seed failing at pool startup: silent.
- JobWatcher: added a stuck-job watchdog that, after 15s with no
job available, logs a specific diagnosis (no chain tip yet vs.
RebuildJob() failing) instead of retrying forever in silence.
- QueuePayouts(): a block found with zero recorded shares queued no
payout and logged nothing.
- FlushMaturePayouts(): an unparseable payout address was skipped
with no log.
- Participant miner: the 'waiting for pool notify' stall was only
ever shown as a transient GUI status string, never written to
debug.log. Now logs on a 15s cadence while stalled, and after 120s
proactively drops the tunnel so the existing reconnect-in-10s loop
gets a fresh attempt instead of hanging indefinitely.
- Participant miner: share rejections were only a transient GUI
status string; now logged with the pool's rejection reason.
All edited files verified with g++ -fsyntax-only (gnu++14).
* Fix two bugs causing worker stuck + log bloat
1. Worker stuck asking for work (main.cpp):
The participant miner checked msg["result"].is_boolean() to detect
the mining.subscribe acknowledgement (id==1), but the pool sends
an array ([[["mining.notify",sid],sid,4]]) as per Stratum spec,
not a boolean. is_boolean() always returned false, so gotSubscribeAck
was never set true. This didn't directly block authorize (id==2 path
is independent), but it did mean the status never advanced through
"subscribed", leaving the progress logging misleading. More
importantly, the same wrong assumption meant any pool that also
gated job dispatch on a prior subscribe confirmation would leave the
worker stuck forever with no diagnosis.
Fix: accept any non-null result for id==1 (the array is the
session/extranonce info; its contents don't matter for our use case).
2. Log bloat from ProcessMessages (main.cpp):
ProcessMessages() used a bare printf() which fired unconditionally
for every P2P message received -- the bulk of debug.log was this one
line repeating thousands of times. Convert to LogPrint("net", ...)
so it is gated behind fDebug (off by default) and burst-collapsed
by the dedup introduced in the previous commit.
* Fix log bloat, em-dashes, dead code, status bar, Connected/Detected peers
Log bloat:
- Gate all per-message/per-block/per-loop printf in main.cpp, net.h,
net.cpp behind LogAcceptsCategory("net") -- only visible with -debug.
This includes: received:/sending: hex dumps, got inventory, askfor,
getblocks, getdata, received block, AddToWallet, RelayWalletTransactions,
ProcessMessage FAILED, mining loop N-transactions line, retarget block,
socket recv/send errors. AddToBlockIndex new best and ProcessBlock
ACCEPTED intentionally left always-on (one line per block).
- PoolParticipantMiner: status -> now LogPrint(worker) not printf.
Em-dashes:
- Replace all UTF-8 em-dashes (0xE2 0x80 0x94) with ASCII '--' in
every source file (rpc.cpp, gui.cpp, main.cpp, main_gui.cpp, net.cpp,
nostr.cpp, util.cpp, btfrv.cpp, db.cpp, db.h, main.h, headers.h,
headers_core.h, util.h). 40 occurrences across 14 files.
Dead code:
- Remove unused gJobSeq atomic from rpc.cpp (written, never read).
- Fix minerReward: was computed but payout loop used blockReward directly
(operator fee was advertised but never deducted). Now payout correctly
uses minerReward = blockReward * (1 - feeFrac).
Status bar / peers:
- Add GetDiscoveredPeerCount() to nostr.cpp/nostr.h (reads g_btfPeers
size under cs_btfPeers).
- Replace scrollable inline status text with DrawStatusBar(): a separate
ImGui overlay window pinned to bottom of screen (NoScrollbar,
NoBringToDisplayFront). Labels changed from 'Peers N' to
'Connected peers: N Detected peers: N'.
* Rewrite gui.cpp: proper fixed status bar, scrollable content child
Layout architecture:
- DrawMainWindow() covers the full display width, height minus STATUS_BAR_H
(22px). NoScrollbar + NoScrollWithMouse on the outer window so the
chrome (menu, address/balance strip, telemetry strip, separators) never
scrolls.
- A BeginChild("##scroll") fills the remaining height after the fixed
chrome, sized to GetContentRegionAvail().y -- which is already the space
between the cursor and the bottom of the outer window. All mode-specific
content (transaction table, participant pool list, operator workers/payouts)
lives inside this child. Content that overflows gets a scrollbar; nothing
bleeds past the outer window edge.
- DrawStatusBar() is a separate overlay window pinned to the bottom 22px,
drawn last so it is always on top. NoBringToDisplayFront is NOT used here
-- drawing last is sufficient and avoids interaction issues with dialogs.
- Status bar background is fully opaque (not alpha-blended) so it never
shows content behind it.
Labels and content:
- 'Peers N' replaced with 'Connected peers: N Detected peers: N'
using GetDiscoveredPeerCount().
- Mining indicator: '* Mode' instead of the 'filled circle' Unicode glyph
that renders as boxes on some Linux font configs.
- FmtAge used consistently; 'blocks deep' shortened to 'N deep' in tx list.
- Options dialog: em-dashes removed from RadioButton labels (ASCII '--').
- Removed usedY dead variable (GetCursorPosY() result was unused).
* Fix blocking bugs, dead code, pool/participant flow correctness
Bugs fixed:
StratumRecvLine blocking (main.cpp):
The participant miner's recv loop called recv() with no timeout after
select() said the socket was readable. If the server sent a partial
line (split across two TCP segments), the second recv() would block
indefinitely waiting for the rest -- the classic 'stuck asking for
work' hang even though the pool was alive and had sent data.
Fix: single non-blocking recv() pass (MSG_DONTWAIT on Linux,
FIONREAD+recv on Windows). If no complete line is in the buffer yet,
return false; the outer select() loop retries when more data arrives.
BroadcastJob holding mutex during send() (rpc.cpp):
BroadcastJob() held gMinersMutex across all SendLine() calls.
send() can block if a slow miner's TCP receive buffer is full.
With the mutex held, MinerThreadFn could not acquire it to remove a
disconnecting miner, causing a deadlock-adjacent stall: broadcast
blocked on slow miner, disconnect blocked on broadcast, new jobs
not delivered. Fix: snapshot {fd, authorised, address, extranonce2}
under the mutex, release it, then send to each fd outside the lock.
Dead code removed (rpc.cpp):
- gRpcUser, gRpcPassword: defined, never read or set by anything.
- dHashesPerSec, nHPSTimerStart: leftover Bitcoin 0.1 stubs, unused.
- IsInitialBlockDownload(): stub that returned pindexBest==NULL,
never called anywhere in this codebase.
- struct Share: declared with three fields, never instantiated.
- Duplicate comment block before payout section.
Other cleanup (main.cpp):
- STUCK_NO_JOB_TIMEOUT_SECS moved from inside the loop body to file
scope (was 'static const' inside a loop -- correct but confusing).
- pblock->print() in solo miner gated behind LogAcceptsCategory("net")
-- was unconditionally dumping a full block to stdout on every solve.
- ERROR in BitcoinMiner printf converted to LogPrint("net").
* Pool/participant: fix recv blocking, BroadcastJob mutex, listener spawn, escaped em-dash
main.cpp -- StratumRecvLine rewrite:
The previous version called recv() in a while loop -- if select() said
the socket was readable but only a partial line had arrived, the second
recv() call would block indefinitely waiting for the rest of the line.
This was the root cause of the recurring stuck-worker bug.
Rewrite: single non-blocking recv (MSG_DONTWAIT on Linux, FIONREAD+recv
on Windows). If no complete newline is in the buffer yet, return false
immediately; the outer 50ms select() loop retries when more data arrives.
This means the hash loop always stays responsive regardless of network
fragmentation.
rpc.cpp -- BroadcastJob mutex held across send():
BroadcastJob held gMinersMutex for the entire duration of sending jobs
to all connected miners. send() can block if a slow miner's TCP buffer
is full. With the mutex held, MinerThreadFn could not acquire it to
clean up a disconnecting miner, stalling the whole broadcast.
Fix: snapshot {fd, authorised, address, extranonce2} under the mutex,
release it, then send to each fd without holding any lock.
rpc.cpp -- LaunchThread failure not handled in BtfStratumAcceptOneFn:
The replacement listener thread was launched via LaunchThread() which
wraps _beginthread() but discarded the return value. If _beginthread
fails (thread limit, OOM), the listener count stayed at 1 with no thread
behind it -- that relay would get no new listener until the 30s outer
loop in BtfStratumAcceptFn fired. Fix: call _beginthread() directly,
check the return value, decrement and log on failure.
rpc.cpp -- escaped em-dash in log string:
BtfStratumAcceptFn had a literal \xe2\x80\x94 in a LogPrint format
string that survived the em-dash sweep (it was source-encoded, not a
UTF-8 byte in the file). Replaced with '--'.
* Rewrite pool server: single event loop, no per-miner threads
Replace the old architecture (one thread per connected miner + separate
JobWatcher thread + BtfStratumAcceptFn thread with listener count tracking)
with a clean two-layer design:
Accept layer (one thread per relay, AcceptOneFn):
Each relay gets exactly one persistent thread. It loops: calls blocking
RvServiceRegister (waits for a miner to knock), then BtfServiceWrap
(echan handshake), then pushes the finished plain socket to a small
lock-protected queue, then loops immediately to accept the next miner.
No listener count tracking. No replacement thread spawning. No races
around spawn failure. The relay always has exactly one pending accept.
Event loop (PoolEventLoop, runs on the ThreadRPCServer thread):
Single thread owns all connected miner sockets. Each tick:
1. Drain the accept queue -- pick up any newly-connected miners.
2. Check for a new block -- rebuild job and broadcast if so.
3. No-job watchdog -- log and retry if job is missing too long.
4. Periodic job refresh -- restamp nTime every 60s.
5. select() over all miner fds with 50ms timeout.
6. For each readable fd: non-blocking RecvLine, drain all complete
lines, dispatch each to HandleLine, collect dead fds.
7. Refresh GUI stats snapshot every 2s.
Because the event loop is the only writer to gMiners, no mutex is needed
on the miner list. The only mutexes remaining are:
- gStatsMutex: guards the GUI-visible snapshot (StatsSnapshot,
gWorkerStats, gPayoutViews) written by the event loop and read by
GUI threads via GetPoolOperatorStats/GetPoolWorkerStats/GetPendingPayouts.
- gReadyMutex: guards the accept queue between AcceptOneFn threads
and the event loop.
Other improvements in this rewrite:
- gPendingPayouts is event-loop-only (no mutex needed).
- Round share tracking lives in the event loop (roundShareCount map +
roundShareTotal counter) -- zeroed on block found, no per-miner stat
mutex needed mid-round.
- QueuePayouts takes the share snapshot as a parameter instead of
reaching into gMinerStats under a separate mutex.
- BroadcastJob no longer snapshots miners to avoid a mutex -- it runs
directly in the event loop with exclusive access to gMiners.
- AcceptOneFn backs off 30s on relay registration failure instead of
returning -- one relay going down doesn't kill the accept thread.
- No goto. No per-thread dynamic allocation for listener contexts on
each accepted connection.
* Thorough logging and correctness fixes across the full pool/participant flow
Every failure point in the operator->miner->payout chain now logs exactly
what broke, with enough context to diagnose without a debugger.
Operator side (rpc.cpp):
RebuildJob: logs 'no chain tip (pindexBest is null)' on failure instead
of silently returning false.
HandleLine:
- JSON parse errors logged with the raw line (first 80 chars)
- subscribe: logs fd and session id
- authorize: logs address, worker, and whether the address is valid
- submit: logs stale-job case with submitted vs current job id;
logs high-hash with actual hash and target values;
logs valid shares with nonce and running session count;
logs BLOCK CANDIDATE separately from BLOCK ACCEPTED/REJECTED so
a ProcessBlock failure is unambiguous
- unknown method logged (reveals miner software quirks)
- reply() logs send failures with method name
QueuePayouts: logs total payout amount, miner count, fee applied, and
how many miners were below dust threshold.
FlushMaturePayouts: logs 'insufficient funds or wallet locked?' on
SendMoney failure instead of just 'FAILED'.
BroadcastJob: logs job id and miner count on every broadcast.
Event loop: logs when each new miner socket enters the loop; logs
disconnect with session and round share counts.
No-job watchdog: RebuildJob retried every 5s (not every 50ms tick),
eliminating cs_main hammering when there is no chain tip.
AcceptLoopFn: re-polls BtfAllRelays every 60s so relays discovered
after startup are picked up; logs each new relay thread spawned.
AcceptOneFn: logs relay name on start/stop; logs BtfServiceWrap
(echan handshake) failure separately from RvServiceRegister failure.
RecvLine: disconnected out-param distinguishes 'socket closed' from
'no complete line yet', preventing incorrect disconnection on partial recv.
Participant side (main.cpp):
StratumRecvLine: same disconnected out-param fix as RecvLine in rpc.cpp.
Previously a false return could mean either 'disconnect' or 'no data yet',
causing spurious reconnects when a partial line arrived.
PoolParticipantMiner rewritten with step-by-step logging:
Step 1 resolve: logs specific failure (no descriptor vs bad format)
Step 2 tunnel: logs relay and failure reason
Step 3 address: warns explicitly if payout address cannot be determined
Step 4 subscribe/authorize: logs each send and each response
Step 5 RandomX: logs VM mode
Step 6 main loop:
- malformed mining.notify (wrong param count or short header/target)
logged with sizes instead of silently ignored
- authorize rejected: logs reason, does not disconnect (some pools
reject bad addresses but still send work)
- share accepted/rejected: logs share id and rejection reason
- unknown server method logged
- pool disconnect logged with handshake state at that moment
(gotSubscribeAck, gotAuthorizeAck, haveJob) so you know exactly
how far the connection got
- nonce space exhausted logged (was silent before)
- on exit: logs total hashes, shares sent, shares accepted
* Fix build error: ImGuiWindowFlags_NoBringToDisplayFront -> NoBringToFrontOnFocus
NoBringToDisplayFront does not exist in this version of ImGui (v1.89.x).
The correct flag is ImGuiWindowFlags_NoBringToFrontOnFocus (bit 13).
Used in two places: ##statusbar and ##main windows.
Also remove dead variable usedY (GetCursorPosY() result assigned but
never read) and replace the confusing 'ImGuiWindowFlags_NoScrollbar * 0'
expression (evaluates to 0 but looks like a mistake) with a plain 0 in
the BeginChild call.
* Remove duplicate status logging from SetParticipantMiningStatus
Every transition in PoolParticipantMiner already has a direct LogPrint
with full context (addresses, job ids, nonces, error reasons). The
LogPrint inside SetParticipantMiningStatus was firing on top of those,
producing redundant 'PoolParticipantMiner: status -> ...' lines after
every meaningful log entry. SetParticipantMiningStatus now just updates
the GUI string silently.
* Implement vardiff: stable share rate instead of fixed difficulty
Previously the pool set shareTarget = blockTarget >> 8 for every miner
forever. This meant at low network difficulty a fast miner found shares
every few nonces (dozens per second), while at high difficulty a slow
miner might go hours without one -- both are bad for fair payout
accounting and pool stability.
Pool (rpc.cpp):
Per-miner difficulty that adjusts to target one share every 15 seconds.
After every 5 shares, check actual seconds/share vs target. If the
ratio is off by more than 50% (VARDIFF_RETARGET_RATIO), recompute:
new_difficulty = current_difficulty / (actual_secs_per_share / target)
i.e. shares coming too fast -> difficulty goes up; too slow -> down.
Only apply if the change exceeds 10% to avoid jitter.
Difficulty is bounded to [0.001, 1e12].
Each miner gets their own shareTarget derived from their difficulty via
ShareTargetFromDifficulty(), which maps Stratum difficulty to a uint256
target using the Bitcoin difficulty-1 base (0x1d00ffff compact).
New difficulty is sent to the miner via mining.set_difficulty.
Participant (main.cpp):
Apply the received difficulty value from mining.set_difficulty.
Previously the handler was a no-op ('acknowledge only'); now it parses
the difficulty double and logs the change. The share target used for
comparison comes from the pool's mining.notify p[3] field, which the
pool updates after each vardiff adjustment -- this is the standard
approach and matches how cpuminer, xmrig etc. handle it.
With vardiff, share rate is roughly constant regardless of miner speed
or network difficulty. Payouts are proportional to shares submitted,
so each miner's payout accurately reflects their CPU contribution.
* Fix share target mismatch: pool rejected shares it should have accepted
Root cause: the pool put job.shareTarget (from blockTarget<<8, a fixed
per-job value) into mining.notify p[3], but checked m->shareTarget
(from ShareTargetFromDifficulty, the vardiff value) on submit. With
difficulty=1.0, ShareTargetFromDifficulty gives a much HARDER target
than blockTarget<<8, so shares that the participant thought were valid
(passed the easier notify target) were rejected by the pool (failed the
harder difficulty-1 target).
Fix: MakeNotifyParams now takes the per-miner shareTarget as an explicit
parameter. SendJob passes m->shareTarget. Both sides -- the participant
reading p[3] and HandleLine checking m->shareTarget on submit -- now use
exactly the same value. Single source of truth.
Cleanup:
- Remove shareTarget field from StratumJob (was only set, never read
after the MakeNotifyParams call was fixed).
- Remove MakeShareTarget() function (was only used to populate that
now-dead field).
- Participant set_difficulty log: 'initial' vs 'updated' instead of
'unchanged' (first connect sends 1.0 which matches the default 1.0,
so it was misleadingly logged as 'unchanged').
* Fix double broadcast, correct block reward calculation, improve payout log
Double broadcast (event loop vs HandleLine):
When a pool miner finds a block, HandleLine calls RebuildJob() +
BroadcastJob(true) immediately. The event loop then detects
nBestHeight changed (step 2) and was calling RebuildJob() +
BroadcastJob(true) again -- miners received duplicate job broadcasts.
Fix: step 2 now checks gHaveJob before rebuilding. If HandleLine
already built and broadcast a job for this height (gHaveJob=true),
step 2 just logs 'job already broadcast by submit handler' and skips.
If the block came from an external peer (not our miner), gHaveJob is
false and step 2 handles it as before.
Block reward calculation in QueuePayouts:
Was calling CBlock().GetBlockValue(0) on an empty block object.
GetBlockValue uses nBestHeight globally, which by the time QueuePayouts
runs has already been incremented to the height AFTER the found block.
Fix: compute subsidy directly as '50 COIN >> (blockHeight / 210000)'
using the actual found height. At current scale this makes no difference
(same subsidy for thousands of blocks) but is now correct by construction.
Payout log now shows:
- Full block subsidy
- Miner share after fee
- Operator fee amount in BTF (not just percentage)
- Number of recipients, maturity height, dust-filtered count
Example: '[payout] block 2096: subsidy 50.00 BTF, miners get 49.00
(fee 2.00% = 1.00 to operator), 3 recipients, matures height 2196'
* Fix 6 bugs found in full flow re-read
Bug 1 -- Double broadcast (re-applied fix that was lost):
When HandleLine accepted a block it called RebuildJob()+BroadcastJob(true).
The event loop then detected nBestHeight changed and did it again. Miners
received two back-to-back mining.notify clean=true, wasting their work.
Fix: step 2 of event loop checks gHaveJob before rebuilding. If HandleLine
already built a fresh job (gHaveJob=true), step 2 just logs and skips.
Bug 3 -- Disconnect log showed global totals, not session totals:
gParticipantSharesSent/Accepted accumulate since node start, not since the
last connect. Added sessionSharesSent/sessionSharesAccepted local counters
that reset each time PoolParticipantMiner() is called.
Bug 4 -- Dead function ParticipantShareTargetFromDiff (main.cpp):
Defined but never called anywhere. Removed.
Bug 5 -- Invalid-address miner shares silently burned:
If a miner authorized with an invalid address, their share of the block
reward was neither paid out nor logged -- it just vanished. QueuePayouts
now validates addresses up front, logs the forfeited amount explicitly
('X BTF forfeited (bad address)'), and excludes those entries so the
payout log is accurate.
Bug 6 -- SendMoney called from event loop (stalls miner I/O):
FlushMaturePayouts called SendMoney which takes cs_main. If the P2P
thread held cs_main (processing a block), the event loop would stall
for seconds, causing all connected miners to time out waiting for
responses. Fix: FlushMaturePayouts now only moves matured entries to
gPayoutExecQueue (fast, no locks except a brief mutex). A new
PayoutThreadFn thread drains the queue and calls SendMoney
independently, so the event loop is never blocked by wallet I/O.
Bug 7 -- Dead goto + dead HandleLine return-false path:
HandleLine always returns true; the goto next_miner path was never
reached. Removed goto and simplified the select drain loop.
* Fix share-target scale mismatch and dead pool hashrate reporting
- ShareTargetFromDifficulty() based Stratum difficulty 1 on Bitcoin's
SHA256 diff-1 constant (0x1d00ffff, ~2^32 hashes/share) instead of
this chain's actual PoW floor (bnProofOfWorkLimit = ~uint256(0)>>12,
~2^12 hashes/share). That's a ~2^20 (~1,048,576x) mismatch: at the
supposedly-easy difficulty of 1.0, workers were being asked for
shares roughly a million times harder than the chain's own easiest
target. Vardiff never gets a chance to correct this because it only
retargets after real shares come in, and none were coming in. Fixed
by basing the diff-1 reference on bnProofOfWorkLimit instead.
- Pool hashrate was never computed anywhere despite the data (per-miner
difficulty + share timing) being available: ThreadBtfPoolAnnouncer in
nostr.cpp hardcoded ann.hashRate = 0.0, PoolWorkerStatView had no
hashrate field, and the GUI workers table had no column for it.
Added an EWMA hashrate estimate per miner (updated on each accepted
share from expected-hashes-at-difficulty / time-since-last-share),
threaded it through GetPoolOperatorStats/GetPoolWorkerStats, the pool
announcement payload, and the GUI (new Hashrate column + totals in
the telemetry line).
Verified: full clean build via src/Makefile (secp256k1 + RandomX deps
built from source, g++ 13, all four touched files compile and link).
* Remove stale wxWidgets-era build files and orphaned assets
These predate the ImGui/GLFW migration (83af065) and are dead weight:
- src/makefile.linux, build-linux.sh: superseded by src/Makefile +
root Makefile (make linux). The old ones reference obj/uibase.o and
obj/ui.o, which don't exist -- they never worked post-migration.
- src/makefile.vc: MSVC build for the old wx UI (obj\ui.obj,
obj\uibase.obj, obj\irc.obj) -- irc.cpp doesn't even exist anymore,
and nothing invokes this makefile. Windows builds go through
makefile.mingw via the root Makefile's target.
- src/ui.rc, src/uiproject.fbp: wxFormBuilder project/resource files
for the old wx UI. The current Windows resource file is bitflash.rc.
- src/rc/{addressbook,send}*.bmp, check.ico, bitcoin_bc_original.ico:
icons used only by the old wx UI. bitcoin.ico (used by bitflash.rc)
is kept.
Verified nothing includes uibase.h/ui.h/irc.h anywhere, and did a full
clean rebuild afterward (secp256k1 + RandomX from source, src/Makefile)
to confirm removal doesn't break the real build.
* Mine with RandomX fast (2GB dataset) mode instead of light mode forever
RandomXInitDataset() was fully implemented and ready to switch mining
into fast mode, but nothing ever called it -- not in PoolParticipantMiner()
(pool worker), not in the solo BitcoinMiner() loop. RandomXCreateMinerVM()
falls back to light mode whenever the dataset isn't built, so every miner,
participant or solo, has always been hashing in light/cache-only mode.
Light mode exists for cheap single-hash verification (which is why the
pool's own share verification correctly stays on it) -- it recomputes
dataset items on the fly for every hash, which is dramatically slower
than fast mode's direct dataset lookups. Combined with the share-target
scale bug fixed in e87a905, this compounds into exactly the symptom
reported: a participant sitting there hashing for many minutes per share
instead of the vardiff-targeted ~15s, even after the target fix landed.
Fix: build the ~2GB dataset once (multi-threaded, sized to
hardware_concurrency) before mining starts, in both the participant and
solo paths, so RandomXCreateMinerVM() actually gets to use fast mode.
Verified: full clean rebuild via src/Makefile.
* Add vardiff stall watchdog: force difficulty down on zero shares, not just share count
ShareTargetFromDifficulty already fixed the diff-1 scale mismatch (chain
floor vs Bitcoin's SHA256 diff-1), so difficulty 1 correctly means ~4096
expected hashes on this chain, not ~2^32. That part was already correct.
But the normal vardiff retarget in HandleLine only runs after
VARDIFF_SHARES_PER_CHECK (5) shares arrive. A miner that never finds even
one share -- e.g. stuck in RandomX light/cache mode instead of the fast
2GB dataset, or just slower hardware than the target assumes -- never
reaches that check, so it can sit at its initial/current difficulty
indefinitely with vardiff having no samples to act on.
Add a time-based fallback in PoolEventLoop: if a miner's vardiff window
has been open for VARDIFF_STALL_SECS (90s, 6x the 15s target) with zero
shares, force difficulty down by VARDIFF_STALL_CUT (4x) regardless of
share count, and re-notify via mining.set_difficulty. Once a miner is
already at VARDIFF_MIN, log it plainly as a hashrate/connectivity issue
instead of a difficulty issue, so it's not confused with the same-looking
but different problem this was actually fixed for before.
* Fix PrintException rethrow killing the whole node on any transient exception
The uploaded participant log shows the real bug behind the mining stalls:
the node process itself was dying and restarting from scratch every
minute or two ('UNKNOWN EXCEPTION ... in ThreadMessageHandler()' /
'ThreadSocketHandler()'), wiping the RandomX dataset/cache and losing all
progress toward a share -- not a difficulty problem.
Every background thread (ThreadMessageHandler, ThreadSocketHandler,
ThreadBtfConnect, the Nostr helpers) wraps its real work in a
'try { Foo2() } catch(...) { log, sleep 5s, loop again }' self-healing
retry pattern specifically so one bad exception doesn't take the whole
node down. But PrintException(), called from inside every one of those
catch blocks, ended with 'if (!fShutdown) throw;' -- re-throwing the just-
caught exception during ordinary operation. Nothing above the retry loop
catches it, so it escapes the thread and std::terminate()s the entire
process on the very first transient error (a flaky relay connect, a
malformed message, whatever). The comment's intent (suppress noise while
already shutting down) was inverted from what the surrounding retry loops
actually need.
Removed the rethrow. Threads now log the exception and let their own
retry loop handle it, as designed.
* Push a fresh job immediately after any difficulty change, not just on next broadcast
mining.set_difficulty alone doesn't change what the worker actually hashes
against -- the participant only reads its share target from
mining.notify's params[3] (see main.cpp's mining.notify handler); the
set_difficulty message just updates a locally-logged currentDifficulty
value. So both the normal vardiff retarget (in HandleLine, on 5 shares)
and the new stall watchdog were updating m->shareTarget and telling the
worker about the new difficulty, but the worker kept hashing against its
old target until the next natural job broadcast -- a new block, or the
60s trickle refresh. That could delay any difficulty correction by up to
a minute for no reason.
Call SendJob(m, false) right after each set_difficulty send so the
updated shareTarget reaches the worker in the very next tick instead of
waiting on the next unrelated broadcast.
* Build RandomX dataset before stratum handshake, not after
Worker was sending mining.subscribe/authorize, then blocking for the
duration of the synchronous ~2GB RandomX dataset build, and only then
reading the socket. The pool's replies (subscribe ack, authorize ack,
set_difficulty, first job) sat unread in the OS receive buffer for that
entire window -- nothing was lost, but with the build taking anywhere
from seconds to a couple minutes depending on core count, the worker
looked and logged as permanently stuck on 'waiting for pool response'
when it just wasn't listening yet.
Move the dataset build ahead of the subscribe/authorize sends so the
socket is actively being read the moment the pool replies. Also add
missing RandomXDestroyMinerVM cleanup on the two early-return paths
that now happen after VM creation.
* Reset round shares on externally-found blocks, not just our own
Round shares were only cleared in HandleLine() after our own miner's
block got accepted. When the chain height advanced because someone
else found the block, the main loop rebuilt the job but left
roundShareCount/roundShareTotal/Miner::roundShares untouched, so they
kept accumulating across rounds instead of resetting.
* Show pool hashrate in participant view; show immature coinbase value with maturity countdown
* Persist pending payouts to disk -- survive operator restarts
* Fix log escapes, stall timeout 90->45s, save-before-exec ordering, Immature tx status
* Sort transactions by nTimeReceived not chain median time -- new txs always on top
* Sort txs by chain depth not nTimeReceived -- survives resync, unconfirmed always first
* Fix orphan block false-positive payout; meaningful tx status instead of raw depth
* Show Orphaned status in red for orphaned coinbases; distinguish from genuinely unconfirmed
---------
Co-authored-by: Vic-Nas <nasci.victorio@gmail.com>
Co-authored-by: Bitflash Pool Dev <pool@bitflash>
Co-authored-by: Victorio Nascimento <155357371+Vic-Nas@users.noreply.github.com>
Co-authored-by: Claude <claude@anthropic.com>
|
|
* Add Linux GUI installer via Wine
* Fix launcher creation and add BTF_RELAY env var support
* Pool server, Linux GUI, unified binary, README update
* Root Makefile: make linux / make windows, cleaner README
* Remove headless.cpp
* Fix XPM color count mismatch causing toolbar image errors on Linux
* XPM icons: same code on Windows and Linux, no ifdefs around image loading
* Fix GTK icon crash and Linux printf format
* btfrv: 10s socket timeout on relay connects
* Faster peer discovery: more dials per pass, shorter backoff and sleep
* Parallel btf peer connections
* Fix GetMyExternalIP - was disabled, now uses modern IP echo services
* Fix tunnel disconnect on idle and restore external IP detection
* Filter stale Nostr descriptors
* Sort peers by freshness, fresh first
* Pool: add .btf Stratum listener — miners connect via rendezvous, no port forwarding needed
* btf pool: self-replacing listeners, one per relay, no queuing between miners
* Mining modes: solo/operator/participant — pool server on-demand, participant Stratum client
* Status bar: show external IP and .btf address
* README: clean rewrite, concept-first, minimal
* make linux produces AppImage
* make linux produces AppImage
* Pool dashboard: connect instructions, tabs, share bars
* Perf fixes: GetMyExternalIP threaded, btf listener counted
* Perf fixes: GetMyExternalIP threaded, btf listener counted
* Fix pool payouts: defer until coinbase maturity
* Pool: clean mode switching, IP warning
* Pool: clean mode switching
* Fix gPoolRunning: remove static
* AppImage: use real icon from bitcoin.ico
* Fix exit crash; Mining menu
* Fix exit crash; Mining menu
* Fix exit crash; Mining menu
* Fix exit crash; Mining menu
* Fix exit crash; Mining menu
* Windows build: use mingw32-make
* Windows build: simpler make invocation
* Add deps-windows target
* Windows build fixes
* Windows build fixes
* Windows makefile: fix obj dir
* Windows: output to repo root
* Windows: bundle DLLs into zip
* Replace wxWidgets with ImGui+GLFW: same build on Linux and Windows
* Fix %I64d that slipped back through rebase (main.h, net.h)
* MinGW: add _mkdir shim for non-MSVC Windows; fix %I64d
* Fix gPoolRunning extern outside ifndef _WIN32
* Remove -lboost_system: boost is header-only in this codebase
* Status bar: copy buttons for IP and .btf address
* Windows: add app icon via windres, strip wx stuff from rc file
* Clean GUI: single Options menu item, bigger font, better spacing, cleaner options dialog; update README
* GUI: embed Roboto font at 16px for clear readability on all platforms
* Windows: windres depends on obj dir target
* Fix balance hidden off-screen: put on own line below address bar
* Windows: inline mkdir in every rule, remove dir target that confused make
* Windows: move all target to top so it's the default rule
* Fix pool server not starting: always start when operator selected, re-sync dialog state on open
* Write debug.log to data directory (AppData/Roaming/Bitflash on Windows)
* Participant miner: use wallet address as username so payouts go to wallet
* Add .gitattributes to lock LF line endings, prevent CRLF drift
* Refactor pool mining to btf-only transport
* Persist pool settings and startup wiring
* Prune stale peers and expose pool queries
* Fix build regressions and add CLI mode/arg parity
* Fix rpc miner stats declaration order
* Centralize pool runtime extern declarations
* Separate operator/participant dashboards and restore peer stats
* Add live telemetry strip and clearer per-mode runtime metrics
* Reduce mining dashboard clutter and remove manual pool entry
* Count only authorized pool miners
* Seed initial pool job on startup
* Use a real share target for pool work
* Switch pool accounting to round-based shares
* Fix nostr pool announcer stats call sites
* Fix pool stats definition order
* Improve pool worker status display
* Add mutex include for participant status
* Replace ImClamp in dashboard layout
* Clarify participant work status
* Log participant status transitions
* Reduce peer discovery log noise
* Add participant status badge
* Track participant auth acknowledgements
* Log and validate pool job delivery
---------
Co-authored-by: Bitflash Pool Dev <pool@bitflash>
|
|
Relay operators announce their relay with -announcerelay=host:port; every node
discovers announced relays (kind 38502) and adds them to its failover list, so
volunteer relays join with no maintainer edit or approval. Anti-spam cap +
validation; relays are trustless (echan makes a bad relay unable to read/MITM).
README: automatic relay-volunteer guide, 'don't let the network die' section.
|
|
Seed list of 3 geographically-spread relays (Sao Paulo, New Jersey, Almaty);
nodes register at the first reachable one and fail over if it is attacked or
goes down, so no single relay IP is a point of failure. Relay installer now
applies iptables connection rate-limits to soak up single-source floods.
|
|
Satoshi's bitcoin-0.1.0 reborn with RandomX CPU proof of work, anonymous .btf
addressing over Nostr with CGNAT rendezvous, automatic peer discovery, a
headless Linux node, and Windows GUI wallet. Fair launch, no premine, 21M cap.
|